Leveraging Computational Storage for Cost Efficiency: TCO Case Study

This presentation will analyze how computational storage, particularly data reduction techniques such as compression, can significantly enhance effective capacity and proportionally decrease the Total Cost of Ownership (TCO) per Terabyte of storage. While these techniques can be accomplished using open-source software on CPUs, the integration of these processes directly into the storage hardware offers distinct advantages.

Performance is an equally important factor when considering compression, as there is usually a trade-off between Input/Output Operations Per Second (IOPS) and/or CPU utilization. For instance, many popular SDS can only enable compression when using (SSDs). While the benefits of computational storage can be readily demonstrated in synthetic workloads and raw disk I/O, it can be more challenging to illustrate these advantages when dealing with filesystems or larger scale-out systems.

TCO reductions achievable through computational storage can be clearly demonstrated when considering reductions in CPU usage and potential for server consolidation. To achieve equivalent performance TCO with CPU-based compression, significantly more resources would be required, making computational storage an attractive solution for cost-effective and efficient data management.

The Storage Networking Industry Association (SNIA) TCO model, which includes a field for compression, will serve as our primary analytical framework.

Download PDF