2010 SDC Agenda Abstracts



ADVANCES IN HARDWARE


Single Root I/O Virtualization (SR-IOV) and iSCSI – Uncompromised Performance for Virtual Server Environments

Leonid Grossman, Vice President, Software Engineering, Exar

Abstract

Achieving high throughput with iSCSI on Virtual Machines (VMs) has proven to be difficult, as iSCSI protocol overhead is compounded by the cost of software IOV. Newer 10GbE NICs with hardware IOV support eliminate most of the software I/O virtualization overhead cost, allowing iSCSI software to run on VMs with better performance. Upcoming hypervisor support for SR IOV is facilitating a wider adoption of direct I/O from a Guest, allowing  iSCSI software to achieve the performance typically seen in non-virtualized environments. Next generation of iSCSI hardware is likely to combine the best features (but not the current limitations) of hardware IOV NICs and iSCSI HBAs.

Status of Clustered CIFS Using SambaStatus of Clustered CIFS Using Samba Learning Objectives

  • Best practices for implementing virtualization in an iSCSI environment
  • Learn about new advanced I/O virtualization technologies (SR-IOV)
  • Update on trends in server and storage virtualization


SAS Technical Update and Advanced Connectivity Roadmap

Marty Czekalski, Vice President, SCSI Trade Association (STA)
Jay Neer, Member Representative, SCSI Trade Association (STA)

Abstract

Serial Attached SCSI (SAS) architecture and technology is growing within enterprise environments. This session will detail the SAS roadmap, connection schemes, how these connections evolved, and the various components that are creating a vibrant infrastructure for expansion. We will introduce the SAS Advanced Connectivity Roadmap and update on the direction for the next generation of SAS (12Gb/s SAS). Finally, there will be a preview of what these advances mean for SAS growth within the enterprise.

Learning Objectives

  • The details necessary to assemble and deliver compelling SAS solutions
  • About the SAS Advanced Connectivity Roadmap
  • What SAS Connectivity Management means
  • What Future Technology Generations of SAS will mean to the Enterprise
  • Why SAS Advanced Connectivity is important for extending SAS’ reach into emerging cloud computing, large data centers and enterprise storage opportunities


Out-of-Band Recovery Mechanism for I/O module using Alternate Control Path

Mayank Saxena, Software Engineer, NetApp

Abstract

Alternate Control Path is a subsystem which provides non-disruptive recovery method to recover I/O module, which is responsible for serving data back-n-forth between disks and Operating System, independent of data path. Whole recovery mechanism is automatic and proactive i.e. when I/O interruption is detected and I/O module/s is/are identified as source of problem then ACP recovery mechanism get triggered to recover shelf module to start serving data without any I/O interruption. Provides capability to reset the I/O module to clear the error conditions without any significant delay to I/Os. It uses Ethernet infrastructure to provide control capability for all I/O modules connected in a particular system configuration.

Learning Objectives

  • Automatic recovery using ACP reduces percentage of problematic I/O modules which encountered unrecoverable errors but behaves fine when brought offline for diagnosis.
  • For transient errors where shelf modules are not responding and corrected by using ACP recovery wont need manual intervention to reseat the shelf module.
  • This improves storage subsystem’s availability for serving data.


BEST PRACTICES IN STORAGE TESTING


BUGS...Catch them before they catch you! A Survey of Storage Test Tools and Industry Best Practices

Peter Murray, Product Manager, SwiftTest
Alan Newman, Vice President Marketing SwiftTest

Abstract

Test tool vendors are in the unique position of understanding how vendors test their products.  Traditional Network Storage testing relies on benchmarks and home-grown tools.  Storage developers often rely on home-grown tools for performance testing, home-grown tools and standard clients for validation testing and benchmarks for comparison with other products.  Recently, commercial-grade tools have appeared on the market, with vendors claiming to address performance and scalability testing.    In this presentation we will share what we have learned (without names!), including how freeware, benchmarking tools, homegrown tools and commercial tools are used today.  We will also present emerging best practices    Each test tool type offers value, but no single tool has yet emerged that meets all storage testing requirements.  This is because multiple approaches are required to characterize device- and system-level performance. Benchmarks are useful for testing certain performance characteristics of NAS systems, but do not accurately mimic real user behavior.  Homegrown tools can be expensive, painful to maintain and are inflexible

  • You will understand how freeware, benchmarking tools, homegrown Storage Testing tools and commercial Storage Testing tools are used today 
  • You will learn about the leading Storage Networking commercial tools currently available
  • You will learn about Storage Testing best practices as learned from our customer base


Error Injection on an Emulated Virtual SCSI Disk

Pulkit Tanwar, Software Engineer, Wipro Technologies

Abstract

Error injection is a mechanism of inducing errors in an otherwise error free device or procedure. Its application mainly lies in the field of testing of error handling code. For storage solutions, error injection techniques are based around the usage of a hardware setup to inject or emulate errors on a storage device.  This paper discusses a method of injecting errors on a SCSI disk emulator without the use of any hardware error injection setup. It also describes a way of exposing the error injection setup to iSCSI interface and other transports like SAS and FCP using SCST which is a SCSI target subsystem implementation for Linux.

  • Get introduced to SCSI testing and debugging
  • Introduction to various SCSI errors


BIRDS OF A FEATHER


BoF: Data and Storage Management for Quantum Computing

Lei Liu, Principal Engineer, Oracle

Abstract

Quantum computing is a mature technology. This session discusses how to design and implement quantum data representation, management and distribution.

Learning Objectives

  • How to represent quantum data
  • How to manage quantum data
  • How to distribute quantum data
  • How to encode quantum data
  • How about security


BoF: Cloud Storage vs Object Storage

Jerome Lecat, CEO Scality

Abstract

Cloud Storage vs. Object Storage – an industry map With over a hundred companies and research groups positioning themselves as Cloud Storage players, it is becoming difficult to understand who does what! The BOF session will propose a categorization of the players, discuss which categories are ideal for different use cases and also review which product developers should look for when integrating their application with Cloud Storage.



BoF: Membrane: Operating System Support for Restartable File Systems

Abhishek Rajimwale, University of Wisconsin

Abstract

Membrane is a set of changes to the operating system to support restartable file systems. Membrane allows an operating system to tolerate a broad class of file system failures and does so while remaining transparent to running applications; upon failure, the file system restarts, its state is restored, and pending application requests are serviced as if no failure had occurred. Membrane provides transparent recovery through a lightweight logging and checkpoint infrastructure, and includes novel techniques to improve performance and correctness of its fault-anticipation and recovery machinery. We tested Membrane with ext2, ext3, and VFAT. Through experimentation, we show that Membrane induces little performance overhead and can tolerate a wide range of file system crashes. More critically, Membrane does so with little or no change to existing file systems thus improving robustness to crashes without mandating intrusive changes to existing file-system code.



BoF: quFiles: The Right File at the Right Time

Kaushik Veeraraghavan, University of Michigan

Abstract

A quFile is a unifying abstraction that simplifies data management by encapsulating different physical repre- sentations of the same logical data. Similar to a quBit (quantum bit), the particular representation of the logi- cal data displayed by a quFile is not determined until the moment it is needed. The representation returned by a quFile is specified by a data-specific policy that can take into account context such as the application requesting the data, the device on which data is accessed, screen size, and battery status. We demonstrate the general- ity of the quFile abstraction by using it to implement six case studies: resource management, copy-on-write versioning, data redaction, resource-aware directories, application-aware adaptation, and platform-specific en- coding. Most quFile policies were expressed using less than one hundred lines of code. Our experimental results show that, with caching and other performance optimiza- tions, quFiles add less than 1% overhead to application- level file system benchmarks.



Long Term Information Retention

Sam Fineberg

Abstract

As more of the world’s information is digital throughout its entire lifecycle, we are faced with the age old issues of record keeping and preservation, applied to devices and formats that were never intended to last.  Long-term digital information suffers from issues that didn’t exist in short-term or paper world, such as media and format obsolescence, bit-rot, and loss of metadata.  The SNIA Long Term Retention (LTR) TWG has taken the lead on this issue for the storage industry.   Working with key stakeholders in the preservation field, the LTR TWG is developing the Self-contained Information Retention Format (SIRF).  SIRF is a low-level data format that enables applications to interpret stored data, no matter what application stored it originally.

Learning Objectives

  • Recognize the difficulties in long-term digital preservation, and how they differ from short-term.
  • Learn best practices and techniques to mitigate the risks inherent in long term preservation, and how those can be applied to today’s storage systems.
  • Identify some of the related standards developed within SNIA and other organizations that are relevant for long term digital preservation. 
  • Discuss the need, use cases, requirements, and proposed architecture of SIRF.


SSD Applications, Workloads & SSS Architectures

Eden Kim, Calypso Systems, Inc.
Dimitri Obukhov, Sandforce

Abstract

Device level synthetic tests per the SNIA PTS provide ordinal rankings of SSDs based on specific access patterns and use cases. What user applications are common, what are the access patterns of those applications and how does one associate a given user application with a set of synthetic device level traits? What use cases are SSDs, HDD/flash hybrids and SSS Cache good for and what does it buy me?
Attendees: Usual suspects
Key Concepts: Examples of user workloads and their synthetic characteristics and when it is better to use flash with HDDs.
Take aways: How much flash I can use depends upon what I am doing. What are the drivers and trends in storage tiering and flash cache?



BLOCK STORAGE


Analyzing Large-scale Network Boot Workload for Windows

Qiang Wang, Development Lead, Microsoft Corporation

Abstract

The presentation covers techniques to measure, visualize and further analyze disk workload for large-scale network boot for Windows. The workload includes both deployment and boot of Microsoft Windows for diskless clients using Microsoft iSCSI Software Target as an example.  The techniques are broadly applicable to any block storage technology.  The presentation also illustrates translation from workload analysis to software scalability opfization as well as underlying hardware configuration guidelines, including clusters.

  • Server scalability considerations for large-scale network boot 
  • Workload measurement and visualization 
  • Workload-specific optimization in software and hardware


Software FCoE - A Case Study

Sudheer Nair, Technical Architect, Patni Americas, Inc.
Manu Gupta, Technical Architect, Patni Americas, Inc.

Abstract

FCoE (Fiber Channel over Ethernet) is an emerging data center technology, the benefits of which include reduced data center management efforts, cost and power consumption amongst others. We are planning to present an introduction to FCoE followed by discussions related to software and hardware based FCOE implementations. We are planning to cover the following as part of the presentation: Examples of FCOE implementation on host followed by comparison of hardware and software implementations.

  • High level design and short comings of OpenFCoE
  • Approach for configuring two servers as FCoE initiator and target using OpenFCoE
  • Also discuss the design approach for one other software based FCOE solution and compare it with OpenFCoE

Learning Objectives

  • Understand differences between hardware & software FCoE implementations 
  • Understand Design of software implementations 
  • Understand comparison of Design approaches (pros & cons)


PCI Express I/O Virtualization Explained

Richard Solomon, Host Interface Architect, LSI

Abstract

The PCI Express I/O Virtualization standards are filled with TLAs (Three Letter Acronyms) – what do they all mean?  Which of the three standards apply to storage developers?  This presentation, by one of the contributors to the specifications, will cover the basics of PCI Express I/O Virtualization as they apply to software developers and system integrators.

Learning Objectives

  • Attendees will gain a basic understanding of PCI Express I/O Virtualization
  • Attendees will learn the differences between various PCI Express function types in a virtualized environment
  • Attendees will learn how to program I/O Virtualization components for best effectiveness


CLOUD


The Evolution of Cloud Storage - From "Disk Drive in the Sky" to "Storage Array in the Sky"

Dan Decasper, Co-founder & CEO, Cirtas
Allen Samuels, Co-founder, Cirtas

Abstract

Widespread enterprise adoption of public cloud storage services requires new storage technologies to be developed that evolve capabilities from today's "disk drive in the sky" with security, performance, and data management issues, to an enterprise-grade "storage array in the sky" that is practically indistinguishable from on-site storage.

Learning Objectives

  • What are the shortcomings of today's cloud storage services that have hampered widespread adoption? 
  • How must existing technologies like automated tiering, caching, and SSDs be adapted to solve these limitations? 
  • How WAN networking technologies will be applied to storage to solve these limitations. 
  • What data processing should take place prior to storing information in the cloud? 
  • What problems can be solved with cloud storage that works like a "storage array in the sky" that cannot be solved with an on-site storage array?


CDMI Specification for Developers

Mark Carlson, Senior Architect, Oracle

Abstract

The Cloud Data Management Interface (CDMI) is an industry standard on it's way to ANSI and ISO ratification. Storage vendors and Cloud providers have started their implementations of the CDMI standard, demonstrating the reality of interoperable cloud storage. This talk will drill down into the developer details of interoperable cloud storage, demonstrating CDMI with examples of the actual messages and message bodies for each of the resources. Familiarity with RESTful web protocols is recommended.


Developing with CDMI

David Slik, Technology Architect, NetApp

Abstract

The SNIA Cloud Data Management Interface (CDMI) provides a powerful yet simple-to-implement interface for accessing cloud storage from your application or web site. This session provides a technical introduction to the standard, and demonstrates simple techniques that can be used to leverage the power of CDMI from within your existing and new cloud-enabled applications.

Learning Objectives

  • Understand CDMI from the perspective of a cloud storage client 
  • See how cloud functionality can easily be added to an existing or new application 
  • Learn techniques to allow cloud storage functionality to be added to Ruby, Javascript, and iPhone OS applications


Developing a Cloud Client

David Moolennar, Vice President West Coast, R2AD, LLC
Michael Behrens, CTO, R2AD, LLC

Abstract

An overview of the development of a cloud client which utilizes both the SNIA Cloud Data Management Interface (CDMI) specification and the OGF Open Cloud Computing Interface (OCCI) specification.

Learning Objectives

  • Understand client side development of cloud specifications 
  • Utilization of JavaFX for client side UI 
  • Lessons learned from interoperability collaboration 
  • Why standardized specifications are important


CDMI Clients

Rich Ramos, Chief Technologist, Consultant

Abstract

This session will cover general aspects of cloud clients as well as specific aspects of developing CDMI clients. the session is intended as a general overview since there are a wide range of clients.  Intended audience should be any interested in development or technical details on CDMI clients.

Learning Objectives

  • CDMI client basics 
  • Developing CDMI clients 
  • Example client demo: iPhone/iPad


The Cloud Data Management Interface Reference Implementation

Scott Baker, Technologist, Consultant

Abstract

The Cloud Data Management Interface(CDMI) SNIA Architecture provides an interoperable and vendor-neutral control and data path management interface for both public and private clouds. The CDMI Reference Implementation provides a basis for vendors to test CDMI implementations for compliance and to test new CDMI clients. This session will provide an overview of the reference implementation including architecture elements, implementation design, code examples, and demonstration.

Learning Objectives

  • Understand key aspects of the CDMI architecture necessary to understand the reference implementation itself. 
  • Learn RESTful principles and how RESTful open source components are leveraged to route RESTful HTTP operations to the CDMI object hierarchy of Containers, Data Objects, Capabilities, and Account Domains. 
  • Learn how the CDMI object hierarchy maps to a simple file system. 
  • Learn how the CDMI Reference Implementation is being leveraged to develop compliant interoperable cloud management clients.
     


Implementing CDMI for Developers

Mark Carlson, Senior Architect, Oracle

Abstract

The Cloud Data Management Interface (CDMI) standard also has a reference implementation written in Java and leveraging the standard JAX-RS Java standard for RESTful web services. This talk will focus on the reference implementation being created by the SNIA Cloud Storage Technical Work Group. We will walk through the Java code and show how to install and build the code, discuss how the various modules are invoked, and talk about ideas for doing your own implementation and extensions.



Cloud Storage - Hands on Lab for Developers (HoLD)

Abstract

The Cloud Hands On Lab for Developers at SDC will include an introduction to CDMI, exploring the current CDMI reference implementation, and an opportunity to write some sample software using CDMI clients.



Cloud Storage - Hands on Lab for Developers (HoLD) - Continued - pending

Abstract

The Cloud Hands On Lab for Developers at SDC will include an introduction to CDMI, exploring the current CDMI reference implementation, and an opportunity to write some sample software using CDMI clients.



Facilitating the Open Source Community Transition to Cloud Storage

Giorgio Regni, CTO, Scality
Brad King, Director Customer Architectures, Scality

Abstract

Object based REST access is becoming the norm for cloud storage services. Unfortunately most services are using their own flavors of REST and applications have to be adapted one by one and for each service. A set of high performance open source libraries and adapters (NFS, FUSE, CIFS, SMB) to facilitate the transition away from block based storage is proposed.

  • high performance REST storage 
  • emulating common storage protocols 
  • adapting common applications to cloud storage 
  • interoperability between cloud providers 
  • security aspects of cloud storage


Leveraging Cloud Storage through Mobile Applications Using Mezeo Cloud Storage REST APIs

John Eastman, CTO, Mezeo Software

Abstract

With the emergence of Cloud Storage, a new service-oriented architecture storage infrastructure has emerged allowing programmatic access to storage.   While today’s mobile workforce relies more on mobile devices to connect them to their business, there are still limitations in terms of storage and security.  Cloud Storage provides an ideal solution.  Join us as we discuss the use case and demonstrate how to use Mezeo Cloud Storage Platform APIs to access storage on a mobile device.

Learning Objectives

  • Understand why organizations are turning to Cloud Storage enabled by APIs.
  • Learn how to use the Mezeo Cloud Storage Platform APIs to access Cloud Storage. 
  • See how to leverage the APIs to support mobile applications.


Use Case: Storage Management and Monitoring Interface via REST

Butch Clark, Engineering Manager / Team Lead, Xiotech

Abstract

Using a RESTful interface to both the end storage and the monitoring and reporting server, Xiotech has been able to dramatically reduce time to market while simultaneously increasing quality.  This session will cover the overall architecture, rapid development process used, issues encountered, benefits of using REST as an interface, tools used, sample code, and demonstration.

Learning Objectives

  • Learn the fundamentals of a RESTful architecture 
  • Understand how you can reduce time to market while increasing quality by using REST, with a rapid development methodology 
  • Benefits of starting or ending with XSD’s (or if you use them at all!) 
  • Learn from our experience what tools and techniques will get you the fastest results


Using SMI-S with the Cloud Data Management Interface

Scott Baker, Technologist, Consultant

Abstract

The Cloud Data Management Interface(CDMI) SNIA Architecture provides an interoperable and vendor-neutral control and data path management interface for both public and private clouds. The Storage Management Initiative Specification(SMI-S) provides an interoperable and vendor-neutral interface for management of SAN systems. SMI-S is the ideal solution to provide the control path implementation for CDMI implementations that are being used to export legacy protocols such as iSCSI and CIFS. This session will provide an overview of such a solution including standard to standard object mapping, code examples, and demonstration.

Learning Objectives

  • Understand key aspects of the CDMI  and SMI-S architectures necessary to understand how the two can work together. Learn how the CDMI object hierarchy can map to an SMI-S profile object model. 
  • Learn RESTful principles and how RESTful open source components can be leveraged to route RESTful HTTP operations to an SMI-S client middle-layer.


CDMI and Cloud Federation

David Slik, Technology Architect, NetApp

Abstract

As Cloud Storage matures, both as a technology and as an industry, individual storage clouds will begin to connect together, share capabilities, and focus on different areas of specialization. These advanced use cases are predicated upon a standard way to represent objects stored in the cloud, and a standard way to communicate with clouds. This session discusses how the SNIA Cloud Data Management Interface (CDMI) provides this standard foundation, and enables cloud federation, peering, and specialization.

Learning Objectives

  • Understand emerging cloud federation, peering, and specialization use cases 
  • Learn how the CDMI standard facilitates these use cases 
  • Explore open challenges in cloud federation, and approaches to create interoperable clouds


Hadoop File System Internals

Dhruba Borthakur, Project Lead, Apache Hadoop DFS, Facebook

Abstract

This talk describes the general design and some recent developments in the design of the Hadoop Distributed File System (HDFS). It talks about the three typical use cases of HDFS.  As HDFS continues to move from a batch workload to a more real-time workload, this talk describes the changes done to HDFS to remove single-points-of-failure. This talk also describes the scale and growth of some of the largest HDFS clusters that are being deployed these days.

Learning Objectives

  • Cloud storage 
  • Challenges with High Availability 
  • Large scale fault tolerance


Designing Secure Storage for the Cloud

Jesus Molina, Security Researcher, Fujitsu America

Abstract

Cloud computing services must be secured to be successful. Data protection is uppermost. This talk will address how to design storage that embeds trust into hardware and how to develop storage systems that secure data in cloud applications. How to apply industry standards will be addressed as will the variations in storage types.

Learning Objectives

  • How trusted storage specifications impact cloud services 
  • How to design storage devices that use industry specifications 
  • Role of trust in securing the cloud


Cloud Security Using Information Dispersal

Julie Bellanca, Founder and Director, Cleversafe

Jason Resch, Sr. Software Engineer, Cleversafe

Abstract

Cloud storage is hampered by security concerns that information dispersal can address. Information Dispersal divides data into fragments and stores the fragments on separate hardware appliances across multiple data centers. Since data is stored in fragments, full copies of files do not exist on any single piece of hardware, offering a paradigm shift for data security in the cloud.

Learning Objectives

  • Learn how information dispersal works with an M of N architecture 
  • Learn how storing fragments addresses security challenges in the cloud 
  • Learn how data can be encrypted and key management can be handled securely 
  • Learn how object storage scales to multi-petabytes required for the cloud 
  • Learn how geographic redundancy and availability are achieved without the overhead of replication


An Enterprise Messaging Solution using Integrated Open Source Software

Prateek Sinha, Storage Solution Developer, Tata Consultancy Services

Abstract

Organizations are increasingly focusing on deploying Open source products in their IT and Business Enablement environments. Several of these stand alone open source products however lack some critical features essential for successful deployment and usage. This drawback can be overcome by a combination of Open Source products working collaboratively in an integrated environment. As an example, let us consider Zimbra Collaboration Suite Community Edition, an Open Source messaging solution, which lacks certain features in the area of Data Protection like Online backup, recovery and seamless Disaster Recovery. We conducted a Proof of Concept by integrating Open Storage, rsync and Heartbeat with Zimbra and were able to address most of the gap feature requirements. This paper articulates the background, framework and approach for the Proof of Concept. We would be demonstrating performance related data for the complete solution.

Learning Objectives

  • Understanding of Zimbra Mail Server, Open Solaris, Rsync, Heartbeat to develop integrated solutions
  • Understanding how current business requirements like Backup/Recovery, Disaster Recovery, Archival can be met using Open Source Software’s
  • Understanding how integration of open source products can help in closing the gap between open source software and proprietary software
  • Understanding of cost benefit accrued while deploying such solutions


CIFS/SMB/SMB2


SMB2 Meets Linux:  Developing the SMB2 and CIFS Kernel Clients for Linux

Steve French, IBM - Linux Technology Center

Abstract

The new SMB2 kernel client for Linux is turning out to be well suited for Linux, providing better data integrity, performance and features.   This presentation will review and demonstrate the progress on the Linux kernel client for SMB2, as well as discuss the current status of the Linux cifs client, and will show areas where the SMB2 protocol is well suited to Linux, and areas where the proposed SMB2 Unix Extensions will make it even better.

Learning Objectives

  • Show how useful the SMB2 Unix Extensions are 
  • Show where SMB2 can be useful for Linux, and how to use it 
  • Show higher reliability available with SMB2 
  • Show progress on Linux cifs client and recent improvements in it too 


MS-FSA: Describing Wire Visible Behavior of Microsoft File Systems

Neal Christiansen, MS-FSA: Describing Wire Visible Behavior of Microsoft File Systems 

Abstract

LMS-FSA is a new algorithms document developed by Microsoft’s file system team to describe wire visible behavior of Microsoft file systems.  This document defines an abstract data model and then provides a detailed algorithmic description of how Microsoft file systems operate.  MS-FSA is referenced by the Common Internet File System (CIFS) Protocol Specification [MS-CIFS], the Server Message Block (SMB) Protocol Specification [MS-SMB], and the Server Message Block (SMB) Version 2 Protocol Specification [MS-SMB2].

Learning Objectives

  • Understand why MS-FSA was developed
  • Understand what is covered by MS-FSA
  • Understand what is not covered by MS-FSA
  • Provide examples of how the document may be used


Are We There Yet?: The Long and Winding Road to SMB/CIFS Specifications

Christopher R. Hertel, Storage Architect and CIFS Geek, ubiqx Consulting, Inc.
José Rivera, SA3G, ubiqx Consulting, Inc.

Abstract

It took two years to produce a new set of SMB/CIFS specifications, which now total approximately 1K pages.  Now that [MS-CIFS] and [MS-SMB] are in maintenance mode, how can the CIFS community make the best use of them?

Learning Objectives

  • How to read the new SMB/CIFS specifications, and related documents. 
  • How to provide feedback to Microsoft and the community. 
  • How long-standing questions have been resolved. 
  • Where to go from here (the future of SMB/CIFS). 


CIFS Map of Solaris

Jiri Sasek, Ing, Oracle
Vladimir Marek, Ing, Oracle

Abstract

Both production Solaris releases 9 and 10 are bundling Samba. Development release of Solaris Nevada from which the OpenSolaris is branched releases also the internal CIFS server and smbfs (CIFS client file-system). This presentation aims to map the authentication algorithms, encryptions, data signing mechanisms... available on these different CIFS platforms.



A Comparison Between the Samba 3 and Likewise Lwiod SMB File Servers

Steven Danneman, Software Development Engineer II, Isilon Systems

Abstract

The Samba 3 smbd file server is the most predominantly deployed open source SMB server in the world.  Lwiod is an open source SMB file server developed by Likewise Software.  Both provide Windows file sharing functionality on Unix operating systems. In this presentation I'll compare and contrast the overall architecture and feature sets of these two servers from both an administrator and developer's perspective.  Attendees should gain an understanding of the primary benefits and drawbacks of each implementation.

Learning Objectives

  • The major architectural differences between the two servers.
  • How threading model fundamentally affects the implementation of key SMB features on POSIX operating systems.
  • The compromises in correctness and scalability each server must make to implement NTFS semantics on top of POSIX. An understanding of all major features provided by an enterprise SMB server stack.


Status of Clustered CIFS Using Samba

Volker Lendecke, Samba Team /SerNet

Abstract

The last 12 months have seen huge steps of clustered Samba towards an enterprise-level product. In particular the usability and stability of the network-wide persistent databases has been improved a lot, making the registry cluster configuration rock-solid. Other advances of clustered Samba include a much improved single-sign on with AD using winbind and better integration with advanced GPFS features.



IPv6 Enabling CIFS/SMB Applications

David Holder, Co-founder and Director, Erion

Abstract

Today IPv6 is no longer the protocol of the future, it is being extensively deployed in production networks. The central pool of IPv4 addresses will be exhausted in 2010 to 2011. As a result, Governments and organisations world-wide have been mandating its use for a number of years. As a CIFS/SMB developer, the time has come where you can no longer ignore IPv6. IPv6 is the default protocol in all major operating systems, including Windows Server 2008 and Windows 7. CIFS and Active Directory have had support for IPv6 for a number of years. This paper will address the key issues facing CIFS/SMB developers when migrating their applications to IPv6. It will provide a review of the status of IPv6 in Samba and Windows and describe how CIFS/SMB applications should be migrated to IPv6. This presentation will also include a brief IPv6 API and porting tutorial presenting the key things that developers need to know about IPv6 and the IPv6 socket API in order to effectively use IPv6 in their applications. Dr David Holder is closely involved in the migration of Samba to IPv6. He has worked with IPv6 for well over a decade and is a regular speaker at global IPv6 conferences.

Learning Objectives

  • Understand the support for IPv6 in key CIFS/SMB implementations
  • Appreciate some of the key differences between IPv4 and IPv6
  • Learn the basic principles of migrating applications to IPv6

 



Failover and Resource Motion Delay in the Microsoft Windows Clustered Fileserver

Dan Lovinger, Senior Software Design Engineer, Microsoft

Abstract

This talk will examine the behavior of the Windows SMB2 clustered fileserver during the process of moving service between cluster nodes. We’ll examine the differences between planned and unplanned motion, both when unloaded as well as under significant external stress. Publicly available data sources on the Windows platform will be discussed and used to build visualizations of the processes involved, and highlight design considerations for solution and system implementers.    To accurately model your client and server implementations, increased attention to real-world usage patterns is required. The File Server Capacity Tool (FSCT) Home Folder scenarios will be used to stress the system. This tool allows measurement of server capacity as well as identification of bottlenecks using a standard, well-characterized workload.

Learning Objectives

  • Examine the building blocks of the Windows Clustered Fileserver, and the possible ways to scale them 
  • Understand the scenarios that make up the FSCT Home Folder workload" 
  • Identify design considerations at scale, and how load can impact system behavior


Implementing SMB2 Within Samba

Jeremy Allison, Engineer Google

Abstract

Samba now implements SMB2. This presentation will explain the code changes that were needed, and how the new features for SMB2 were built on top of the existing stable codebase.

Learning Objectives

  • Samba internals 
  • SMB2 protocol 
  • SMB2 implementation details
     


Introduction to the CIFS/SMB/SMB2 Plugfest

Abstract

Abstract Pending



Interoperability Tools for CIFS/SMB/SMB2

Paul Long, Technical Evangelist, Microsoft

Abstract

Interoperating with Windows can be  significantly simplified by learning how to use the tools and resources available to you.  Tools such as Network Monitor, Test Suites, Protocol Documentation and Spec Explorer can help identify interoperability problems and assist in testing implementations.  Resources such as leveraging protocol documentation support and  attending PlugFests are invaluable for insuring your software will interoperate properly with Windows, in particular CIFS/SMB/SMB2 systems.  This presentation will detail the available tools and resources to help you understand the offering.

Learning Objectives

  • Why read the documentation when you have Network Monitor? 
  • Using Test Suites and Spec Explorer to improve interoperability 
  • Find Protocol Documentation from a Network Trace 
  • How to use Unified Messaging (ETW) and NetSh to gather traces 
  • Understanding other resources like PlugFest and Documentation Support


SMB Traffic Analyzer - Monitor the Data Flow of Samba Servers

Holger Hetterich, Level 3 Support Engineer / Samba Developer, Novell

Abstract

SMB Traffic Analyzer is a project with the goal make it possible to monitor and statistically view the data flow on Samba Servers. Rather than a network capturing tool, Traffic Analyzer utilizes the Virtual File System Layer of Samba, and produces a SQL enabled storage, making it easy for the users to generate statistics in general. Furthermore, client programs are being developed, structured on the SQL database, which produce full reports of the data flow and even monitor data flow in real-time.

Learning Objective

  • Using the Samba Virtual File System Layer for Applications


Evaluating SMB2 Performance for Home Directory Workloads

David Kruse, Principal Developer Lead
Dan Lovinger, Sr Software Design Engineer, Microsoft

Abstract

Understanding performance analysis of the home directory workload on file servers can be a complex task in terms of modeling the workload, understanding the bottlenecks, and potentially changing the protocol to improve performance. Utilizing the File Server Capacity Tool (FSCT) from Microsoft to create a realistic workload, this talk will examine the home directory workload in detail to better identify issues for efficient optimization of your client and server implementations.  The talk will provide increased insight into the client workloads over the SMB2 protocol, while also providing pointers on how to gather relevant data while running under load.  It will also discuss the events used for performance analysis on Windows Server 2008 R2, and then look to how the data gathered can be used to create theoretical models of the effects of future protocol improvements.  Finally, we’ll look at how these events could theoretically be exposed to individual clients to allow fine-grained scenario performance evaluation for both enterprise and branch-office workloads.

Learning Objectives

  • Examine a system for gathering, evaluating, and visualizing performance data across a large number of active connections as is common in the home directory workload. 
  • Understand how the data can be used to create a theoretical model to extrapolate potential performance or scalability gains from future innovations. 
  • Discuss how the core architecture can be extended to allow individual clients to profile their scenarios across a variety of server implementations.


Scaling Likewise-CIFS Beyond 50K Concurrent Connections on a Single Node

Gerald Carter, Senior Software Design Engineer, Likewise Software

Abstract

Much effort has been invested recently in clustered SMB/CIFS solutions, however many environments still require high connection scalability from a single host or device.  This session will present the architecture used by Likewise-CIFS, the SMB file server component of the Likewise Open project,  to achieve more than 50,000 simultaneous client connections hosted by a single 4-way SMP host configured with 8 Gb of RAM.  Attendees will also be introduced the freely available connection stress utility developed as part of Likewise Open.

Learning Objectives

  • Understand the Likewise-CIFS threading model 
  • Understand Likewise-CIFS connection, session, and file handle management 
  • Learn how to use the Likewise-CIFS connection stress testing client tool


Embedded Private Cloud Using Samba and Ceph

Kai Blin, Samba Team

Abstract

Thanks to the low power consumption of ARM systems, it is possible to connect enterprise-grade SATA drives to the cloud with a peak power consumption of around 10 Watts.  Using the Open Source technologies of clustered Samba and Ceph on top of these embedded system it's now possible to create flexible storage solutions that can be installed right at the SOHO, with a CIFS-based data access for cross-platform file sharing.

Learning Objectives

  • Using clustered Samba to provide distributed CIFS file sharing 
  • Using the Ceph distributed file system for Samba clustering


Identity Mapping for Windows+UNIX Environments

Jordan Brown, Principal Software Engineer, Oracle

Abstract

The Solaris Operating System includes native support for SMB server and client operations, with full integration as the goal, including seamless interoperation with local and NFS operations.  One interesting aspect of this integration is identity mapping - ensuring that a user gets the same access rights regardless of access point: local application, NFS or SMB.

Learning Objectives

  • SIDs and UIDs - the fundamental disconnect between Windows and UNIX 
  • Mapping schemes:  rules, directory data, ephemeral IDs. 
  • Problems encountered and solutions implemented


Unix Extensions for SMB2 - A Protocol Initiative

Tom Talpey, Architect, Microsoft

Abstract

A new effort is currently under way to specify a set of extensions to the SMB2 protocol to enable support for the Posix file semantics expected by Unix systems.  Years ago, a SNIA effort specified similar extensions to the SMB/CIFS protocol. While incomplete, these extensions were adopted in several implementations  and have provided an important base for use of SMB when communicating between Unix/Posix style systems, while using "normal" SMB to communicate with Windows.   The new effort, originally discussed at last year's 2009 SNIA SMB Plugfest, began in Spring 2010 as an open collaborative one focused on the new SMB2 protocol,  and significant work has been done.    We'll review the scope and goals of the new effort, and make a deep dive into the specifics of what is currently being proposed.  Attendees will leave with a good understanding of what's being done, what to expect the new capabilities to mean to them,  and have a taste for when they'll become reality.



Analyzing Metadata Caching in the Windows SMB2 Client

Mathew George, Sr. Software Development Engineer, Microsoft
David Kruse, Principal Software Development Lead

Abstract

Metadata operations constitute a significant fraction of the network traffic between an SMB2 client and server running a typical file server workload. The simplified command set defined by the SMB2 protocol requires that a client implementation must make judicious use of metadata caching and compounding in order to effectively utilize network bandwidth. In the first part of the talk we will discuss the metadata caching strategies employed by the Windows SMB (should this be “SMB2”?) client implementation to reduce metadata traffic to the server.  Specifically we will look into the 3 different metadata caches – the file-information cache, the directory cache and the file-not-found cache and see how these caches help reduce network round trips. We will also look at various configuration settings to control the sizes and lifetimes of these caches.      The second part of the talk will look at some of the consistency issues that arise due to opportunistic metadata caching. We will also look at potential protocol improvements and show how we can mitigate these problems by improving the consistency and lifetimes of these metadata caches.

Learning Objectives

  • Gain insight into the data and metadata caching policies of the Windows SMB2 client 
  • Learn about client settings that can be used to tune the metadata caching behavior 
  • Understand the benefits these caches provide in different scenarios." 
  • Investigate how the introduction of new protocol capabilities could further refine data and metadata caching


Case Study: MS-DFS - Writing New Drivers to Extend Likewise-CIFS

Gerald Carter, Senior Software Design Engineer, Likewise Software

Abstract

Likewise-CIFS is the SMB file server component of the Likewise Open project's Active Directory integration efforts. The server's loadable driver model provides third-party developers a means to extend current functionality while taking advantage of it's robust, scalable architecture. The existing file sharing features can be extended beyond SMB/CIFS to include NFS and WebDAV while the local POSIX file system driver (FSD) could be replaced with other alternative implementations. This session will use the new MS-DFS driver as an example to teach attendees how to write a new driver for the Likewise-CIFS stack, what name space and resource challenges must be addressed, and how multiple drivers may internally interact with each other.

Learning Objectives

  • Explain the Likewise-CIFS driver architecture
  • Learn how to extend Likewise-CIFS network and filesystem functionality by developing additional drivers
  • Understand how individual drivers interact with each other internally to the Likewise-CIFS IO Manager


FILE SYSTEM


ZFS Advanced

Jeff Bonwick, Vice President and Senior Software Architect, Oracle

Abstract

You already know what ZFS can do, but how does it do it? This talk will get under the hood and describe the core algorithms and design principles. General familiarity with ZFS and C programming recommended.



ZFS What's New

Jeff Bonwick, Vice President and Senior Software Architect, Oracle

Abstract

This talk will describe the newest capabilities in ZFS: synchronous deduplication, encryption, triple-parity RAID-Z, disk/flash hybrid storage pools, and disaster recovery. General familiarity with ZFS helpful but not required.



pNFS Development

Brent Welch, Director Software Architecture, Panasas

Abstract

This talk provides an update on the pNFS development process, with a focus on the adoption of the pNFS code into the mainline  Linux code base.  Discussion will include status of the different pNFS back-end protocols (files, blocks, and objects), as well as testing and performance measurements

Learning Objectives

  • What is pNFS 
  • Benefits of pNFS for scalable performance 
  • Learn about Linux code adoption process Performance and testing status update 
  • How you can use pNFS today
  •  


OpenAFS: Ten Years of Open Source Storage Systems

Jeffrey Altman, Gatekeeper and Elder, OpenAFS

Abstract

OpenAFS was founded on 1 Nov 2000 after the release by IBM of its AFS enterprise storage product as open source.  This talk will summarize ten years of the product's open source development by more than 230 developers and describe where the next ten years might take it.

Learning Objectives

  • What makes an open source project successful? 
  • Why is AFS still used given the resources devoted to CIFS, NFS, NFSv4, Lustre, and other network storage technologies? 
  • What types of organizations depend on AFS?


File Systems: Building Windows File Systems: A Look at the OpenAFS Client for Windows

Peter Scott, Consulting Partner, Kernel Drivers
Jeff Altman, President, Your File System Inc.

Abstract

The name space and data concerns within the AFS world create a unique set of problems when interfacing with the Windows file system API. In particular, how relative and absolute symbolic links, mount points, component substitutions and generally convoluted name space topologies are handled within a Windows file system implementation. The management of data coherency across distributed, dissimilar clients involving server side notification introduces yet more complexity in an already complex world.

Learning Objectives

  • How to handle the nuances of the Windows infrastructure from the outside 
  • Keeping metadata and data coherent within a distributed world involving Windows 
  • Understanding the intricacies of the AFS topology 
  • How the AFS paradigm can be mapped into the Windows world


Ultra High-Speed Transport Over WAN

Michelle Munson, CEO & Co-founder, Aspera

Abstract

The movement of large datasets is a bandwidth-taxing process. But what if there was a way to transfer 20 gigabytes of content in a matter of seconds? By circumventing traditional transfer methods like TCP, FTP and NFS, which all prevent networks from utilizing all their available bandwidth, this fantasy can become a reality. This presentation will address how it’s done and how to overcome the storage bottleneck in the last foot of the transfer

Learning Objectives

  • Understand a new strategy for moving gigabytes of content between workgroups, offices, and continents. 
  • Learn how to optimize performance and bandwidth for large data transfers over modern WANs 
  • Learn how to pinpoint where bottlenecks occur (SAN, NAS, LAN and WAN infrastructure) and the residual impact these bottlenecks have on your network.


Optimizing Disk Layouts for Mixed Sequential Read Workloads

Stephen Daniel, Technical Director, NetApp
Rickard Faith, Sr. Member Technical Staff, NetApp

Abstract

Mixtures of random writes and sequential reads are common in many database and email applications.  Write anywhere file systems such as WAFL have demonstrated their ability to optimize the random writes; however, this write optimization has historically come at the expense of sequential read performance.

This talk will discuss the evolution of sequential read optimization technologies that allow WAFL to simultaneously optimize both random writes and sequential reads.  Finally, using both analytical and measured results the talk will show that write anywhere file systems can outperform traditional fixed layout file systems across a broad range of mixed workloads.



IDA based Virtual Appliance for Secondary Storage Solutions

Giridhar Lakkavalli, General Manager/Technical Architect, MindTree Ltd

Abstract

IDA based Virtual Appliance for Secondary Storage Solutions Typical secondary storage involves multiple storage products or solutions like Backup & Restore, Archival, Disaster Recovery and Replication to achieve data protection, and data availability. But, this approach results in high cost, large amount of storage space (typically 3 to 5 times the size of data to be stored & protected) and high complexity. This presentation focuses on an approach of using an IDA (Information Dispersal Algorithm) based Virtual Appliance (VA) to achieve an effective secondary storage solution that is - cost and space efficient, highly reliable & scalable. This approach also eliminates the need for RAID enabled storage boxes for backup storage as IDA has an inherent advantage of withstanding multiple failures. The lDA appliance approach also handles replication. This appliance approach can be seamlessly integrated into the existing storage infrastructure by SMBs, enterprises to achieve an effective secondary storage solution.

Learning Objectives

  1. Challenges of the traditional secondary storage solutions
  2. IDA based virtual appliance (IDA VA) approach for secondary storage solutions and challenges addressed by IDA VA, and the benefits
  3. Practical environments, and use cases across digital word where IDA VA approach could be leveraged
  4. Comparison of traditional secondary storage solution with IDA VA based secondary storage solution
  5. Future or Roadmap for IDA based storage solution approaches, and what could be the likely paradigm shifts


Issues When Implementating Data Reduction Technologies and Deduplication: Doing Dedup Right

Jeff Tofano, Chief Technical Officer, Quantum Corporation

Abstract

This presentation explores ways data reduction technologies are integrated into traditional filesystems. We'll cover the issues and benefits of integrating each of the major mechanisms - single instancing, compression and dedup into tradition filesystems. We’ll also cover how integrated data reduction schemes add value to traditional data protection, data management and basic data movement tasks.  We'll also explore how dedupe will continue to evolve, becoming an ubiquitous part of file based storage systems, and suited to be a foundation for cloud storage environments.

Learning Objectives

  • What are the issues and benefits when integrating each of the data reduction technologies into a filesystem? 
  • What are the trade-offs customers need to know when evaluating filesystem based reduction mechanisms? 
  • How will dedupe evolve to be a more ubiquitous feature of filesystems and emerging cloud storage infrastructures?


NFS in Windows Platform

Jeff Biseda, Senior Software Development Engineer, Microsoft

Abstract

This presentation will cover an overview of the basic building blocks of NFS implementation in Microsoft Windows and challenges faced in interoperability with NTFS file system, SMB file access protocol and rest of the Windows ecosystem.

Learning Objectives

  • Overview of architecture and design choices made in Windows NFS implementation 
  • Interesting interoperability and compatibility issues with Windows ecosystem 


The Metadata Problem - Understanding the Issues and How to Address Them

Anand Babu Periasamy, CTO & Co-Founder, Gluster

Abstract

Understanding how metadata affects your storage architecture is critical for developing storage that performs well and scales. This presentation will outline major storage metadata models, how different file systems have addressed the problem and techniques and best practices for solving metadata problems for different use cases. The speaker will discuss the pros and cons of centralized, distributed and no-metadata architectures. The presentation will cover object storage, file storage, parallel/clustered file systems, and others as well as an explanation of popular metadata algorithms.

Learning Objectives

  • Understand the different metadata management models 
  • Why metadata management is critical to performance and scale 
  • Techniques for metadata management 
  • Pros/cons of different metadata management models


RESTful Filesystems

James Cain, Principal Software Architect, Quantel Limited

Abstract

Representational State Transfer (REST) has been taking the online world by storm, but most commentators state that HTTP is the only protocol that can implement REST. James will present the key architectural aspects of REST and show how a Virtual File System (VFS) combined with a NAS protocol can support the architectural principles completely. He will then demonstrate a real implementation of the theories on a live system.

Learning Objectives

  • Understanding of REST architectural style. 
  • Understanding of the untapped potential in NAS protocols. 
  • Insight into how a Virtual File System can support a workflow. 
  • Demonstration of working codification of theoretical ideas.


Solving Data Loss in Massive Storage Systems

Jason Resch, Lead Software Engineer, Cleversafe, Inc.
 

Abstract

The problem of data loss becomes harder as storage systems scale and technology advances. Disks become larger and take longer to rebuild; increasing the probability of LSEs and secondary failures. As systems grow to Petabytes, hundreds of individual parity groups exist, each having its own chance to fail. New technologies, beyond RAID 6, are required to address such issues as correlated disk failures, latent sector errors, and silent data corruption, in these massive storage systems.

Learning Objectives

  • The difference between Latent Sector Errors, Silent Errors, Correlated Failures, and Operational Failures
  • Events which lead to data loss in today's systems
  • Why the problem of data loss is becoming more difficult to deal with over time
  • RAID 5 is dead and why RAID 6 won't last for long
  • What technologies can be applied to address this problem


Database Techniques to Manage a Distributed POSIX Namespace and Associated Metadata

David Boomer, Information Architect

Abstract

This presentation discusses the use of distributed relational database techniques to enable exascale file system and HSM metadata management for large-scale Linux clusters. We present metadata management techniques for: scaling a POSIX namespace to a trillion files, file storage allocations information, and file content knowledge over a Linux cluster.   We address major data management challenges while focusing on critical needs to enable scalability for exascale era computing and file storage.   We present test results from a Linux cluster using  High Performance Storage System (HPSS) & DB2 software to show architectural approaches that surpass the file storage limitations of a single system in a cost effective and efficient manner.

Learning Objectives

  • Managing a distributed POSIX namespace 
  • Balancing metadata, storage, and workload across storage servers 
  • File knowledge using a flexible, scalable XML architecture
     


HOT TOPICS IN STORAGE


KMIP - Key Management Interoperability Protocol Implementation Guidance

Gordon Arnold, Senior Technical Staff

Abstract

The new standard for encryption key management called KMIP is being broadly adopted.  Come learn about the implementation best practices for KMIP and the specific profiles related storage implementations.

Learning Objectives

  • What is the state of KMIP implementations 
  • What does it take to enable my encrypting capable storage for KMIP 
  • What are the options for storage client registration 
  • What are the security features for KMIP
  •  



SECURITY


Seven Myths About Storage Encryption

Dmitry Oblukhov, Firmware Security Manager SandForce Inc.

Abstract

The presentation is focused on clarifying popular misconceptions about self-encrypting drives collected for the first year of such products on the open market.  1. ATA security (BIOS password) is good enough  2. Software encryption works better with SSD.  3. Storage device has slow controllers and so has to use weaker algorithms  4. SED introduces a lot of latency  5. TCG Opal is Windows only  6. TCG storage encryption is intended to enforce DRM  7. Storage vendors will keep back doors for failure analysis

Learning Objectives

  • Architecture of the self-encrypting drives 
  • Performance-related concerns 
  • The role of self-encrypting drives in overall security system


Self-Encrypting Hard and SSD Drive Revolution: Managing the Data at Rest Solution

Robert Thibadeau, Senior Vice President and Chief Scientist Wave Systems, Corp. 

Abstract

There is a new breed of self-encrypting lap top hard drives designed to the Trusted Computing Group Opal standard.   US law now provides for safe harbor against Data Breach Notification Laws.  This is causing rapid adoption of corporate lap top encryption technology.  The law stipulates NIST 800-111 which says  “Organizations should use centralized management for all deployments of storage encryption”.  Management of self-encrypting storage devices on corporate lap tops should be factored away from SAN management because the requirements are radically different.

Learning Objectives

  • What does it mean to centrally manage self-encrypting lap top drives? 
  • Why does NIST 800-111 require this? 
  • What is the difference in managing these from encrypting enterprise storage management. 
  • How can developers integrate lap top self-encrypting management into systems management? 
  • Familiarity with the new laws and regulations.

SOLID STATE STORAGE


Storage Acceleration, Driven by Autonomic Software

Sam Siewart, Chief Technology Officer, Atrato, Inc.

Abstract

Data intensive applications are growing in number; a telltale sign of an increasingly complex business environment. The widening gap in client initiator and disk-based storage performance has created an overall inefficiency in the data center. Compounding the need for performance is the explosive growth of data. In this session we will discuss the advent of “performance tiering,” and how it is used to better service the demands of high I/O requests. Learn how new autonomic software aligns the value of data with the value of storage to improve performance/throughput and cost-effectively manage data between tiers. We will also discuss in detail the benefits of a purely autonomic system that dynamically adjusts to changing workloads and data access patterns, in real-time.

Learning Objectives

  • Performance Tiering:  the blending of SSD (Tier 0) and HDD (Tier 1) technology 
  • Next-Generation Autonomics:  dynamic data movement and self-management software 
  • Performance Benchmarks:  essential checklist for autonomic tiering functions


The Write Endurance of SSDs - Is MLC NAND Fit for Data Centers?

Avraham Meir, CTO, Anobit

Abstract

Migrating from Single-Level-Cell (SLC) to Multi-Level-Cell (MLC) NAND reduces SSD bill-of-material cost by a factor of 3. The major limitation for this migration in data center environment is write endurance. This presentation explores the gap in write endurance between MLC NAND and the requirement of data center environment, describes methods to bridge this gap, and presents a new approach based on signal processing, which yields an MLC-based SSD with write endurance of SLC NAND.

Learning Objectives

  • Analyzing the write endurance requirement of data center environment 
  • Analyzing the write endurance of current and future MLC and SLC NAND, and main driving factors 
  • Write amplification and SSD architecture 
  • Data compression and encryption 
  • Memory Signal Processing


Flash and the Architecture of Storage Systems

Avraham Meir, CTO, Anobit

Abstract

The migration from hard disk drives to Solid-State-Drives (SSDs) bears the promise of increasing random access time by multiple orders of magnitude. This presentation explores the impact of this change on optimal architectures of storage systems, including tiering and auto tiering, caching, redundancy and Redundant Array Inexpensive Disks (RAID), and presents new methods for optimal usage of inherent capabilities of NAND Flash, including wear-level mechanisms and new redundancy approaches.

Learning Objectives

  • Tiering and auto-tiering in SSD-based storage systems 
  • Caching and endurance impact on NAND-based systems 
  • Wear-level, hot/cold areas in the SSD and auto-tiering 
  • Classical Redundancy and new approaches to redundancy and RAID in Flash-based systems


Hybrid Storage Architecture Marries Performance and Efficiency

Bill Mottram, VP of Marketing Atrato, Inc.

Abstract

Green data storage aims to lessen the impact on data center power and space consumption. This task is made difficult by the many pervasive challenges data storage presents, in particular, increasing performance and capacity requirements. In this session, we will discuss how a hybrid storage approach can be used to improve data center efficiencies, while increasing capacity and performance of high I/O environments. We will review how intelligent, autonomic software blends the benefits of solid state and spinning disk to improve productivity, optimize service levels, and lower a company’s total operating costs.

Learning Objectives

  • Solid State Disk: the next level of data acceleration 
  • Data Center Economics: matching the value of your data to the cost of storage
  • Autonomic Management: smart software that optimizes data movement between tiers


STANDARDS TO SOLUTIONS


FC-BB-6: New FCoE Functionality

David Peterson, Senior Advisory Engineer, Brocade

Abstract

This presentation provides an overview of current INCITS T11.3 FC-BB-6 working group activities. Items that will be discussed include new terminology, the ENode functional model, VN2VN_Port functionality (i.e., point-to-point and point-to-multipoint), and distributed FCF functionality (i.e., Controlling FCF and FDF).



The Pieces to Managing FCoE

John Crandall, Architect Technology and Standards, Network and Storage Management, Brocade

Abstract

This presentation will identify and provide an overview to the many implementation pieces a client and FCoE switch and CNA vendor may need to consider for managing an FCoE environment including discovery and configuration. These pieces may include standards from the SNIA, DMTF, and T11. The presentation will attempt to  identify the possible gaps in these standards and how they may be addressed and will also examine whether a client can hooks all these pieces together and if so how?

Learning Objectives

  • What are the current and future management standards 
  • How do you hook the pieces together 
  • What are the gaps and how might they be addressed


iSCSI and FCoE Testings: What are the Test Challenges Under the Hood of a 10 Gb iSCSI and a 10G FCoE Storage Array Product Certification?

Mahmoud Jibbe, Distinguished Engineer/Quality Technologist, LSI Corporation

Abstract

The certification of a 10 Gb iSCSI and a 10 Gb FCOE RAID Storage System on the same network elicits a lot of challenges at the development level and the Test / Quality Assurance level. The challenges are due to the fact that both 10 Gb iSCSI and 10Gb FCOE are newly deployed host interfaces in the RAID Storage environment.  As a result, development module test for both interfaces on the same network should be designed very carefully to establish test coverage beyond basic operations with a RAID Storage system, standard RAID testing, or an Ethernet plug fest.  These module level must tackle the timing differences associated with a network being accessed by two different protocols with the following common and different characteristics: :  1. Device discovery for both protocols on the same network   2. 10GB CNA which is common to both protocols  3. 10 GB switch traffic control and congestion which is common for both interfaces   4. Security mechanisms with different Operating systems,   5. Operational parameters associated with I/O retries and recovery  6. Management, Administration, and Integration with Storage products  7. Design For Testability “DFT” mechanisms  8. Debugging, Diagnostics, and problem Isolations  9. IPV4 vs. IPV6

Learning Objectives

  • Testing of service discovery for both protocols on the same network 
  • Testing of 10GB CNA which is common to both protocols 
  • Testing of security mechanisms with different Operating systems 
  • Testing of operational parameters associated with I/O retries, recovery and Management, Administration, and Integration with Storage products, and Debugging, Diagnostics, and problem Isolations
  • There is not much leverage between 1 Gbit and 10 Gbit operation due to differences in chip set and driver firmware.- may be somewhat true for IP level. Again there are considerable differences between 10Gbit iSCSI and 10Gbit FCOE (Chipset, driver, and protocols). Common part is IP layer. There are specific features such as backup, snapshot, remote mirroring, and cluster application compatibility that must be supported by the RAID product and must be verified during the testing of the RAID controller host interface types. As with any inter connected usage, device operation, such as RAID Storage Subsystem, has direct dependency on proper operation of interfaces. 


Tutorial: "Green Storage is About More than kWh!"

SW Worth, Sr. Standards Program Manager, Microsoft

Abstract

At least some of what you 'know' about Green IT is wrong.  Re-thinking your storage technology designs to reflect environmental and economic realities can not only help make your products more competitive, it might even be more effective than putting another bumper sticker on your car.    This talk will cover the three stages of any product life-cycle: ‘birth’ (design/manufacture/install), ‘life’ (operation across all load ranges, including idle and sleep), and ‘death/transfiguration’ (recycle/disposal).  Your decisions as a engineer or developer affect more than just the initial birth-stage; you own some life-cycle responsibility for your creation.  In cost-accounting, these manifest as capital expenses (CapEx) and operational expenses (OpEx), which combine to define TCO.  Can these measures really guide you to the right design decisions?  Do they encourage recycling?  Do they encompass the ‘embedded’ costs of greenhouse gasses (GHGs) – even if your products are built in a country in another hemisphere?  Do they include the costs associated with end-stage disposal of toxic metals and flame-retardant chemicals – the stuff that is in your products to make them more reliable or safer during operation?  Do they provide fair, comprehensive, and easily-understood information for purchasers (customers) to make decisions that are not only good for them, but good for the rest of us as well?  (Bonus question: do government programs like EPA’s ENERGY STAR or the European “Code of Conduct for Data Centres” lead to optimal outcomes?)    One good place to focus is on understanding the differences between Power (Watts) and Energy (Watt-hours), and their respective affects on costs.  The peak power your hardware or software can draw primarily affects CapEx, including data center infrastructure (e.g. UPS, PDU, CRAC, AHU) and the IT gear that it supports (e.g. Servers, Networking, and Storage).  Energy consumption affects OpEx, by way of the monthly bills that the IT department probably never even sees.  But what about all of the capital investment required for the generating stations, transmission lines, and distribution system – how does the utility recover their costs for that?  You’ll learn how real-world billing and charge back works, and what you learn make affect your design and architectural decisions.

Learning Objectives

  • Green: It’s *not* just about electricity consumption!
  • Power and Energy are not the same, and they each affect costs differently. 
  • Embedded’ environmental and infrastructure costs may outweigh IT operational expenses in TCO calculations. 
  • How your designs should change to reflect real-world billing and charge back mechanisms.


Energy Management: The Key to an Effective Storage Consolidation Strategy

Joe Polastre, CTO & Co-founder, Sentilla Corporation

Abstract

An effective storage consolidation strategy is critical for efficient and functional day-to-day operations; however, managers often overlook the most important component of their strategy; energy management.      Managers should consider the following energy management issues when formulating storage consolidation strategy:  o Operational costs and cost impact of storage   o Power and cooling impact of consolidation decisions  o Ensure consolidated storage is appropriately sized for virtual/cloud  infrastructures  o Cost reports to each business line for storage usage

Learning Objectives

  • How to provide cost reports to each line of business for their storage usage. 
  • How to look at the cost impact of storage in terms of used TB/Watt and money used per TB. 
  • How to determine the operational costs of each tier and redundancy of storage. 
  • How to evaluate the power and cooling impact of consolidation decisions. 
  • How to ensure consolidated storage is appropriately sized for virtual/cloud infrastructures.


Process of Generating Power from Hard Disk Vibration in Datacenters

Senthilkumar Vijayakumar, TATA Consultancy Servicesl

Abstract

This talk concentrates on Energy Harvesting. Energy harvesting (also known as power harvesting or energy scavenging) is the process by which energy is derived from external sources, it’s captured, and stored. Energy harvesters currently do not produce sufficient energy to perform mechanical work, but instead provide small amount of power for powering low-energy electronics. Energy harvesting is also motivated by a desire to address the issue of climate change and global warming. Based on proof of concept, we will present a process to sense the vibration of the hard disks in the datacenter, and converting these vibrations to a control signal for pneumatic control. The pneumatic control has piston type of arrangement which produces massive stress on piezoelectric sensor. Piezoelectric sensor which has the ability to transform mechanical strain energy into electrical charge. The power so generated can be stored in a capacitor, super capacitor, or battery so this energy can be used for energy conservation and harvesting. In datacenters Energy harvesting devices can be used for powering the operations of micro electronics, microcontrollers and processors in datacenters..



Introducing and Validating SNIA SSS Performance Test Suite

Esther Spanjer, Director, SSD Technical Marketing, SMART Modular

Abstract

This presentation will provide an overview of SSSI Performance Test Suite (PTS), that was proposed by the Technical Working Group of SSSI. It will walk through the various steps of the proposed benchmark test suite to ensure that reliable, comparable and repeatable performance test data is achieved.     The presentation will include a review of select results of an industry wide Blind Survey conducted by Calypso Systems, Inc in May 2010. This private Blind Survey tested many leading SSD products using the test methodologies and tests recommended by the SNIA SSS TWG Performance Test Specification rev 0.9.

Learning Objectives

  • Get a good understanding of the various parameters that influence the performance characteristic of a solid state drive. Test results will be shared to illustrate the influence of these parameters.
  • Get a full understanding of the proposed SSSI Performance Test Suite. Topics that will be discussed include preconditioning, queue depth, idle tests, random vs. sequential test suites, etc.
  • Provide a step-by-step guidance on how to set up a test benchmark that creates equal results among the various SSS devices.


Trends in Solid State Storage

Jeff Kimmel, Senior Technical Director, NetApp

Abstract

This talk explores trends in solid state storage and their likely impacts on system architectures. Included are SLC/MLC considerations, wear leveling techniques and costs, read and write efficiencies vs. HDD, flash caching layers and their placement in systems and tradeoffs between caching and use of SSDs.



STORAGE MANAGEMENT


SMI-S Over WS Management:  A Progress Report

Josh Cohen, Senior Program Manager, Microsoft
Jim Davis, CTO, WBEM Solutions, Inc.

Abstract

“Web Services for Management” (‘WS-Man’) is the first specification in support of the DMTF initiative to expose CIM resources via a set of Web services protocols.  By identifying a core set of Web service  specifications and usage requirements to expose a common set of operations that are central to all systems management, WS-Management has the ability to:  • DISCOVER the presence of management resources and navigate between them  • GET, PUT, CREATE, and DELETE individual management resources, such as settings and dynamic values  • ENUMERATE the contents of containers and collections, such as large tables and logs  • SUBSCRIBE to events emitted by managed resources  • EXECUTE specific management methods with strongly typed input and output parameters    SMI-S has included the DMTF- standard WS-Man as an optional transport protocol (instead of CIM/XML) for several versions.  Now, SMI-S v1.6 is making that a reality.  How much change will this mean for SMI-S?  Who is doing the Standards work, and when will they be done?    Learn about WS-Man, and why it offers some advantages over CIM/XML.  More importantly, learn about what changes this could mean for your current and future implementations of SMI-S, in both providers and client applications.

Learning Objectives

  • Basic understanding of WS-Management 
  • I like CIM/XML and it works well enough, so why should SMI-S change?  (More importantly, why should I change my implementation?)" 
  • OK, I’m convinced, but who is going to do the work to get WS-Man into SMI-S v1.6?  (And when are they going to finish?)


SMI-S Client and Server Development, from Check Box to Industrial Strength

Steve Peters, Project lead, PMC-Sierra

Abstract

Complying with SNIA's SMI-S must be more than a check box. The presentation will chronicle the development of a SMI-S provider and a web based GUI that manages a RAID card. The talk will include the evolution from simple to a full featured implementation. It will describe how the added features were used to enhance the GUI and the user experience.

Learning Objectives

  • Implementing a simple provider makes it hard for the GUI. 
  • Implementing a provider cache. keeping in sync with lifecycle indications. 
  • Performance improvements using view classes.


Embedded SMI-S Lessons Learned

Mike Lamb, TPC, Storage Resource Management Software, IBM
Art Colvig, Software Engineer, IBM

Abstract

Discussion of the challenges and design decisions involved in embedding SMI-S agents on the TS3500, SVC, DS8000 and XIV devices. 

Learning Objectives

  • CIMOM Selection 
  • Footprint reduction 
  • Porting to a CIMOM to an RTOS 
  • Design tradeoffs 
  • Why embed?


Automatic Generation of SMI Providers

David Dodgson, Software Engineer, Unisys

Abstract

An SMI Provider must support the numerous classes defined as part of the storage profiles.  The Provider is usually in the form of a DLL attached to a CIM Object Manager.   It is used to make the instances of these classes available to the SMI clients to provide management of the storage device.  It is cumbersome to write the code to support all of these classes.  A method is described that will generate the provider code automatically based upon the descriptions in the MOF files.  This is done through additions to the Simple WBEM open source code to provide functions that implement the intrinsic interfaces to store, retrieve, modify, enumerate, and associate the instances through references.  A build procedure is defined that uses these functions to generate provider code for each class, and combine them into a single DLL.

Learning Objectives

  • Describe the additions to Simple WBEM used to provide the intrinsic interfaces. 
  • Describe how the instances are stored and retrieved from memory. 
  • Describe how a fully-formed provider for a class is generated. 
  • Describe how the class providers are combined in a DLL.
     


VIRTUALIZATION


Developing Storage Architects for a Virtual World

Alex Miroshnichenko, CTO, Virsto Software

Abstract

Most storage technologies deployed today were designed based on assumptions that are true for physical servers but false for virtual data centers. The result of this can include is storage sprawl, poor application performance, storage management complexity and excessive storage costs. This session will discuss how hypervisor-based solutions can get to the root of the problem, explore best practices for architecting storage in a virtual world and provide real-life deployment scenarios for improving storage deployment in virtual servers.



"Sparse Storage" for VMs - Often Overlooked Facts

Dilip Naik, Managing Partner, VMUtil Inc.

Abstract

The discussion will explore the details of thin provisioned storage which can be accomplished at the block or file system level (example NTFS sparse files.  The similarities of the problem between thin provisioning and SSD need for trim support will be explored. The discussion will also dive into the implications of how Virtual Machine files are stored, copied, and backed up.  Perform traces will be used to show that thin provisioned storage is not necessarily PCI bus efficient, unless care is taken with the software used to copy and backup VM files.

Learning Objectives

  • Understanding VM file storage optimization 
  • How to evaluate backup software


LSI's VMware vCenter Plug-In: A Study in the Use of Open Source Software

Brian Mason, Staff Software Engineer, LSI
Erik Johannes, Staff Software Engineer, LSI

Abstract

LSI has developed a plug-in to better integrate management of our Storage Array product line with VMware’s vCenter management suite.  The plug-in is a Rich Internet Client that integrates with vCenter. As part of its Raid Development effort, the team has utilized several open source toolkits, including: SBLIM, GWT and Java.  This case study details the architecture that glues the toolkits together and what has worked and not worked.

Learning Objectives

  • Use of open source tools aid in Rapid Development 
  • Use of Web Technologies for Rich User Experience 
  • Architecting for aiding protocol independence and testability
     


Improve Storage Efficiency on Deploying Oracle Databases for Dev/Test in a Virtualized Environment Over NFS

Bikash Roy Choudhury, NetApp

Abstract

Server virtualization technologies provided by VMware vSphere 4 allows you to virtualize database environments and create a base virtual machine that runs an Oracle Database 11g R2 properly installed and tested. This base virtual machine, which can be a golden copy of a  real production Oracle database system, is stored as a set of files in a shared storage system, e.g. NetApp FAS. It is self-contained, including a guest operating system(Linux), Oracle home, and all the data and log partitions. So using RCU(rapid Cloning utility) to clone the virtual machine using NetApp’s storage efficient thin cloning technologies gives you all you need for application development, QA testing, and report generation in a  Dev/test Oracle database 11g R2 setup. This eliminates the challenges of extended system downtime, degrade system performance during the cloning process and also eliminates the use up a large storage space to store each cloned virtual machine.

Learning Objectives

  • Best practices of running Oracle single Instance in a VSphere 4.0 environment 
  • Ease of cloning the Oracle VMs using RCU in a Dev/Test environment is instantaneous and saves a lot of storage space. 
  • Eliminates the need for developers to have physical hosts on which  the database clones need to be mounted and also can be automated using RCU from VMware vCenter server. 
  • To take consistent backup and restore oracle database volumes or files using SnapManager for Oracle 
  • Ways to migrate Oracle DB from existing storage environment to NetApp
     


Storage I/O Control: Proportional Allocation of Shared Storage Resources

Chethan Kumar, Senior Member of Technical Staff, VMware

Abstract

Datacenters have benefited by virtualization which has reduced footprints of datacenters, decreased management cost by consolidating physical resources, eased datacenter management, and offered live migration of computing entities. Virtualization has greatly increased the need for a shared storage infrastructure. Storage systems are accessed by many virtual machines hosted on multiple hosts with varying demands and business priorities. Effective resource management and prioritization among workloads is the key to successful storage consolidation and reduction over-provisioning of storage resources. Existing QoS mechanisms provided by the storage arrays are not sufficient to differentiate I/O priorities at virtual machine level. Research and development at VMware has come up with a way to auto manage and prioritize IO resources among the many workloads sharing common storage. In this talk, I will give an overview of this new concept called Storage IO Control. I will present results of experimental studies conducted to understand the impact of SIOC on application performance. I will also illustrate the dynamic capabilities of SIOC in managing shared I/O resources across an ESX based virtual cluster with performance numbers. Finally, I will present a case study that will underline the need for a fine grained control mechanism for prioritizing I/O resources in a virtual datacenter with shared storage infrastructure and show how SIOC can fill that need

Learning Objectives

  • Virtualization
  • Need for a fine grained control mechanism for prioritizing I/O resources in a virtual datacenter with shared storage infrastructure
  • One of the possible solutions
  • Benefits of the solution


Virtualization Storage Concepts, APIs and Performance Best Practices

Aditya Vasudevan, Technical Architect, Patni Americas Inc

Abstract

Virtualization is a key technology to enabling cloud infrastructure. So we present virtualization-storage concepts such as virtual datastores, thin/thick disks, etc from virtualization stand point, in this case VMware as it’s the market leader. We also present the APIs available to the storage community which can improve IO efficiency by making the storage VM-aware and offloading work to it. Finally we present storage configuration best practices for high performance in VMware platform.

Learning Objectives

  • Learn the storage options and concepts on virtual platform (VMware)
  • Learn about vStorage APIs which allows Storage integration with virtual platform
  • Learn the best practices to configure storage for high performance on a virtual platform