The Myths of All-Flash Storage Consolidation
The business benefits of using flash in storage arrays are not always clearly spelled out by many vendors in the industry. We often see flash discussions that focus on speeds and feeds and how a blazing fast platform will “change your applications” and make everyone happy. Arguments along these lines are often heard from industry veterans; but the core message is that lower response times lead to a happier data center.
Taking the next step
Some companies talk about the fact that their all-flash array (AFA) allows for consolidation of different workloads when the economics of putting all workloads on an AFA cannot be justified. These companies may make the business argument that you can consolidate based on performance. Or they may argue that there is so much horsepower in the said platform, that why not assign many different workloads on the same system?
That argument starts to sound more like a business value argument. However, even a cursory examination would reveal that an effort to assign many workloads on the same system is simply consolidating on the single dimension of “performance.” What about the additional dimensions of capacity, different protocols, all-flash and hybrid pools, different block sizes and varying space-saving technologies? Why would a company not want these additional dimensions of consolidation that can be provided by a flash storage platform?
Multi-dimensional storage consolidation: does the math work?
Instead of the one dimension of “performance” offered by other AFA vendors, we instead suggest that a six-dimensional consolidation platform would be ideal. At this point, we could digress into analogies of string theory and the unification of gravity and quantum mechanics that extend beyond the four common dimensions of the space-time continuum. But we’ll skip string theory for now. However, one small digression is in order, as the core logic is applicable.
The problem with unifying gravity and quantum mechanics can be partially summarized by stating that the math does not work when you are constrained to four dimensions (three for space, and one for time); in fact, the math blows up into infinities.
We need to look at higher dimensions. And when we use high-brow math, guess what? The equations make sense again. The analogy also holds true for flash in storage arrays. The math (from a cost perspective) does not work out nicely when you look at a just one dimension of consolidation (performance). However when we calculate six different ways to consolidate (let’s call them dimensions) the math works and becomes compelling.
What types of consolidation does Tegile offer?
Tegile IntelliFlash arrays provide consolidation across these many dimensions. Let’s say you have various types of pools. One may be all-flash for latency-critical workloads, and another may be a hybrid-based pool optimized for performance and capacity. Another option in this area of consolation is to start with an all-flash pool and add expansion shelves of spinning media (if you have plenty of performance left, but are running out of capacity). Only Tegile offers the flexibility for all-flash arrays and hybrid arrays in the same datacenter.
A Tegile array supports multiple protocols that span block and file simultaneously. You can consolidate file and block on the same array. The shares and LUNs can be built with varying block sizes, thus allowing tight alignment with the application IO request size, and thus consolidating very disparate workloads on the same system.
Also, Tegile offers in-line deduplication and compression all the time, if desired. However for some workloads, such as relational databases where the blocks will not dedupe, a user can selectively turn off dedupe. Here we are consolidating workloads with varying space-saving technologies applied.
Now we may have had too many physics analogies already; but let’s do just one more. Recall that the ultimate goal of science is to “prove” theories via experiments and real-world data. Physicists will collide particles at extraordinarily high speeds to show that building blocks really exist (i.e., Higgs Boson). Likewise, Tegile uses real data from our customers to show that multi-consolidation of workloads is possible, and in fact, optimal, to get the best return on investment from storage arrays. At Tegile, we are so confident about our multi-consolidation benefits (based on seeing the raw data from our growing installed base) that we offer a guarantee. We call it our Flash 5 Guarantee.
The Road to Savings
Companies want their flash investment to make good business sense, and to result in a positive return on investment. So, when you are figuring the math on a potential flash investment, be sure to consider more than just the one dimension of performance. Instead, consider multiple dimensions, including capacity, different protocols, all-flash and hybrid and pools, different block sizes, space-saving technologies — and performance. When you include multiple dimensions to calculate the return on your flash investment, the math will solve nicely.