Historically businesses have been forced to aggregate storage with compute workloads to take advantage of the fastest disks. This led to over-specified systems with unused, poorly-allocated capacity. With the adoption of SSDs, the cost of provisioning aggregated data became even more expensive.
Improvements in interconnect technologies have helped to increase the speed at which data can be accessed over distributed storage. InfiniBand, RDMA and RoCE offer virtually no latency at all (>1 millisecond), allowing for storage to be disaggregated from connected systems.
Disaggregation and edge computing
Much of the industry focus for disaggregation has been in terms of boosting read/write access for computing at the edge – providing real time analytics for IoT applications and the like. However these high speed, high volume operations also require super-fast storage – typically SSDs
As a result, the role of legacy storage has been ignored – which makes sense when OEMs drive the uptake of new interconnect technologies. They want their customers to purchase new hardware after all, which is why the focus of disaggregation is usually on SSDs.
Disaggregation extends to archive data too
The benefits of disaggregation (increased expansion options, lower cost upgrades, greater flexibility) are available for any application however – including archiving. Indeed, disaggregation helps to make archive data more readily available – perfect for use in analytics operations.
Because performance is less of issue with these datasets, your business can use spinning disks to add significant capacity at minimal cost. Further savings can be made by redeploying existing hardware into the disaggregated infrastructure, allowing you to reduce total cost of ownership and increase return on investment.
Obviously your OEM is unlikely to recommend this course of action (there’s no money init for them), but that’s not to say this is not a viable choice for your business. In the age of big data analytics, it may be that redeploying existing assets is exactly what your organization needs.
To discuss your data disaggregation options and how much your business could save, please get in touch.