In “Automated storage tiering options for your environment” on SearchStorage.com, Phil Goodwin does a good job detailing the automated storage tiering segment and establishing two categories, intra-array products and inter-array products. I agree with his assessment that there are a lot of differences in the ways vendors have implemented this technology and that in order to make a choice about which products offer the best fit, users need to evaluate their specific needs based on their environments and workloads, etc. For VARs, I think these two categories of automated storage tiering products offer different opportunities.
Intra-array automated storage tiering products run within an array and essentially virtualize data objects into sub-LUN blocks, or “chunks,” and store them in different areas, or pools, on the array. These chunks are typically created according to appropriate parameters, such as drive type or RAID level. Different vendors use different sizes of chunks, and some even allow users to configure those sizes. The automated tiering process itself involves the transparent movement of these data blocks between the established storage tiers in accordance with established parameters.
Intra-array tiering has the effect of decreasing overall storage costs. As a set-and- forget feature, it runs in the background and through the “magic” of automated storage tiering let users put more top-tier data into the box than would otherwise be possible capacitywise. Typically, these systems put new writes onto the lower tiers and migrate data up only as its access patterns warrant.
At one point, intra-array automated tiering was an enabling technology that drove the value proposition for storage array systems. Compellent was one of the first vendors in this space and essentially built its business on this value proposition. Today, however, many storage systems include automated tiering. While there is a lot of differentiation among intra-array automated storage tiering products, automated tiering is now a more common array function. So instead of being an enabling technology that a VAR could use to drive a deal and displace a competitor, it’s now more of a feature that VARs need to understand so they can sell it, or sell against it.
Inter-array automated storage tiering, on the other hand, allows data to be moved externally, between storage arrays systems of the same type or from different vendors. For VARs, this implementation may bring more opportunity than intra-array automated storage tiering since it offers more flexibility by incorporating existing storage systems. Some products, like F5’s ARX file virtualization solution, are more suited to what could be called “archive tiering,” essentially pulling older, reference data out of primary NAS devices. Other products, like Avere’s FXT , could be called “performance tiering” as they front-end NAS devices and provide a faster storage tier to boost performance, especially as these NAS environments scale.
Inter-array solutions may also appeal to a wider group of potential customers because they’re usually file-based and, unlike intra-array solutions, don’t require the purchase of a new storage array. That said, many disk array vendors that include intra-array AST also support external arrays. Inter-array tiering products can also drive more project-oriented work, meaning implementing systems instead of point products, like a single storage array. A project to implement an automated tiering solution between storage systems can also drag along additional components, like servers, networking and, of course, professional services. From a VAR’s perspective, this is the kind work that pays the bills.
Automated storage tiering and SSD
Due to its costs, the capacity of the typical SSD implementation is very small compared with disk storage. So it follows that getting and keeping the right data on SSDs is one of the key challenges with implementing this technology successfully. The ability of intra-array tiering solutions to move small blocks of data automatically into and out of an SSD tier quickly, and transparently, can be very effective at solving this issue. And, most intra-array tiering solutions support a “tier 0’ comprised of SSDs. For VARs with customers looking at SSD to improve application performance and in the market for a new storage array, automated storage tiering can be a good solution.
For customers not interested in a new storage array, some inter-array automated storage tiering solutions, like Avere’s, can be a good fit for implementing SSD. But a similar technology, caching, may be a good alternative as well. Although, technically, a cache involves copying the most active data sets to a fast storage area--usually SSD or DRAM--and tiering involves moving data between tiers, caching can still provide some of the same benefits as storage tiering. Solutions like Dataram’s XcelaSAN and CacheIQ can increase the performance of an existing storage solution and eliminate inefficient disk practices, resulting in a lower total cost of storage. They can also enable a lower-performing storage array to support applications that would otherwise require more expensive storage, also driving down costs.
Intra-array automated storage tiering solutions are available from most disk array vendors, so VARs should first become familiar with what they may already have access to. It’s not a technology that will drive disk deals by itself, but it’s something that VARs need to understand. For new opportunities, inter-array tiering may present a better opportunity, since these solutions offer more flexibility and file-based support and can lead to more profitable project work. For customers interested in solid-state storage, both intra-array and inter-array solutions can be a good way to implement SSDs.
Eric Slack is a senior analyst with Storage Switzerland.