Storage industry analysts are overwhelmingly identifying hybrid storage systems as storage solutions of choice when compared to all-flash solutions. In this series, we’ll take a look at individual analyst recommendations around hybrid storage arrays, starting with my friend George Crump, respected analyst and owner of Storage Switzerland.
In his analysis entitled Overcoming The Challenges To All-Flash Arrays, George identifies two key advantages that hybrid arrays have over their all-flash brethren and he
George identifies a number of marketing techniques used by all-flash array vendors to improve the perceived bottom line of their solutions. These techniques—such as providing a solutions post-data reduction effective capacity rather than the solution’s pre-reduction raw capacity—generally place the all-flash arrays in a more favorable light than would otherwise be possible, but George points out that it’s not really possible to predict data reduction levels for every customer, so these claims may not hold true for every customer. In other words, the pricing claims made by many all-flash vendors are based as much on hope as they are on previous experience with other customers. Many CIOs are loathe to make decisions based on hope and prefer to make decisions based on hard facts.
George further points out that these same vendors often attempt to make their price comparison claims against the most expensive 15K RPM drives on the market. However, in reality, most customers have a combination of very low cost SATA disks and higher performance 10K RPM and 15K RPM spinning disks in their environments, making the oft-used price comparison dubious at best.
George identifies another potential weakness in the all-flash space as the inability to tier data that doesn’t require constant all-flash storage. This fact limits all-flash arrays to niche use cases, but as all-flash vendors seek to become more broadly used in the data center, George feels that they will need to gain the ability to deeply analyze workload performance and potentially move running workloads to other storage systems in the environment—perhaps to the very disk-based storage systems the all-flash solution was intended to replace. In other words, all-flash arrays will need to learn how to become hybrid ones if they want to succeed in the broader market.
“Hybrid all-flash” storage
In this same article, George postulates that there may be an opportunity for all-flash players if they’re willing to add rotational storage to their solutions. This is the reverse of the traditional hybrid play, which is usually spinning disk with an SSD tier of some kind.
Believe it or not, the solution that George postulates is here today with Tegile’s HA2800 product. The base HA2800 array is an all-flash device with 3.5 TB of raw capacity. However, with the addition of the J2800 expansion chassis, which boasts 72 TB of raw capacity, Tegile’s solution turns hybrid on its head by adding massive capacity to an array that already boasts massive performance. It’s all flash storage with a hybrid twist that simply works.
It all comes down to this: For now, moving completely away from rotational storage simply doesn’t make sense from an economics or a performance perspective. Hybrid arrays provide customers with the ideal blend and balance of cost and performance and do so in a way that enables these customers to solve challenging business problems.