File Management

The Abstracts

High Availability and Disaster Recovery for NAS Systems

As network attached storage has matured, users have entrusted increasingly critical data to it, creating requirements for protection against failures and disasters. This session will present a survey of techniques available for protecting data stored on NAS systems against loss or destruction by threats ranging from hardware and software component failure to accidental or deliberate corruption to disasters that completely incapacitate an entire data center. Backup, RAID and mirroring, system fail-soft, snapshots, continuous and periodic replication, and NAS system clustering will all be discussed. For each technique, the threats against which it protects, the capital and operating costs, and the expected recovery time and recovery point objectives will be presented. The goal for the session is to give students an appreciation for the high availability and disaster protection options available for their NAS-managed data, in order to better equip them to make well-informed decisions when purchasing or defining operating procedures.

Learning Objectives

• Give attendees an appreciation for the differences between different HA and DR techniques for protecting data stored on NAS systems.
• Provide abstract cost: benefit analyses of the various HA and DR techniques to assist in decision making.
• In general, equip attendees with the knowledge to make more informed decisions with respect to data center HA and DR implementation and on-going administration options.

Recent Advances in WAN Acceleration Technologies

Access to data hosted in central data centers has been a continuous challenge for application users in remote branch offices. WAN-related performance problems associated with bandwidth and latency usually pressure IT managers to deploy file and application servers in the branch offices themselves, in order to maintain application performance and end-user productivity. But maintaining and backing-up remote server and storage assets outside of the data center is not only expensive, it also creates significant security risks. This session explores new approaches involving disk-based deduplication, TCP protocol optimization, and application-level chattiness mitigation to address this long-standing productivity vs. cost-efficiency issue. We will compare and contrast the strengths and weaknesses of these new approaches with more traditional methods of addressing this problem.

Learning Objectives

• Examine the pros and cons of compression, caching, and adding WAN bandwidth, and understand why none of these measures adequately solves the essential problem.
• Understand why many applications perform poorly over the WAN in an environment with high latency and/or limited bandwidth.
• Explore how TCP protocol optimization, application-level chattiness mitigation, and new disk-based deduplication approaches that can dramatically improve performance over the WAN and enable greater centralization of storage resources