Comment An analyst tells me Solidfire’s technology and locations both have inherent limitations that will prevent its product business from scaling.
The first point, says the analyst, is that its technology over-optimises for capacity, via deduplication, and it doesn’t have enough resources left to process and manage its metadata. Consequently it can choke on metadata and its interconnect, limiting its scalability, our source claimed. There are also service-level agreement and stability issues, this person has claimed.
With the fall in price of SSDs, via 3D NAND and TLC, the absolute need to deduplicate will diminish and extreme tech gymnastics will no longer be needed, negating Solidfire’s capacity advantage over the next two or three years, said the analyst. Ditto XtremIO’s, in the analyst's view.
This will help all-flash-array suppliers, who have optimised for performance and find that capacity is becoming cheaper.
A second point relates to Solidfire’s Boulder, Colorado, location, which is stuck in the middle, far from the hardware and software engineering recruiting grounds of Silicon Valley and the New York/Boston area, says our source. This will limit Solidfire’s ability to grow, says the analyst, and may mean it has to move either east or south west, which its current staff may not be willing to do.
This may strike you as a contrarian view and it needs a skilled engineering look at Solidfire’s tech to verify the capacity over-optimisation point. The recruitment geography point could be correct and could stop Solidfire becoming a storage super-power on its own. It all makes you stop and think.
Solidfire's director for International Marketing, John Rollason, commented on these points:
- Solidfire's array (and XtremIO's) is capacity-optimized to the point where it chokes on metadata and the interconnect
- This is a question we get from time to time. Truth is that SolidFire has been designed from the ground up to scale-out. We can scale today non-disruptively to 100 nodes and 7.5m guaranteed IOPS to more than 100,000 volumes. SolidFire scales cluster connectivity as the cluster scales out, each node having its own network connectivity. There is no interconnect pinch point in the architecture.
- The Solidfire array has stability issues.
- This sounds more like FUD from a competitor. We are proven in production in the world’s largest, most demanding and visible Enterprise and Service Provider customers around the world like CenturyLink, 1&1, eBay, etc.
- It also has SLA limitations.
- [A] strange question as SolidFire is the only storage platform on the market that allows customers to provide meaningful performance and availability SLAs to their end users / customers. We have had customers swap out an entire cluster (tech refresh or redeployment) without impacting service, availability or performance.
- With flash getting cheaper, due to 3D TLC and so forth, the need for resource-sapping deduplication work will be less important and Solidfire's capacity advantages will be less and less effective in competing with performance-optimised AFAs, such as Pure Storage's, over the next 2-3 years.
- We think organisations will continue to value storage platforms that can optimise for both capacity and performance. For SolidFire de-duplication in particular is a fundamental part of how we write data. It’s not an additional service, limitation, overhead or process, it is simply built into the architecture. This coupled with falling media costs will enable customers to continue to drive down the effective cost of storage. As you know, SolidFire has been shipping TLC for over a year now.
- Solidfire's growth prospects over the longer term will be limited by its Boulder, Co, location, as it doesn't have access to the great mass of HW/SW storage engineers in Silicon Valley and the New York-Boston area.
- Being based out of Boulder has been a significant advantage to this point in SolidFire’s evolution, allowing us to build an incredibly talented and loyal storage engineering team. Many other innovative technology companies are also establishing centres of innovation in the Boulder area.
Basically a denial of all the points our source, who wishes to remain anonymous, has made. Take that, sir! ®