Yesterday as part of the first A3 Tech Live event, I sat through a presentation from Memblaze, a vendor of NVMe and PCIe flash storage products. Although the conversation was good natured and “robust”, I couldn’t help feeling that the company (like so many others in this market) have little to differentiate their products, other than on cost. We discussed features such as performance (throughput) and latency, where a quick comparison showed that Memblaze didn’t really have any significant advantage over competitors such as Samsung, Intel, SanDisk (Fusion-io) and others. Yes, there are marginal differences, but in reality products like PCIe SSD cards are so over-specified that they are unlikely to be fully utilised when used to resolve legacy performance issues and so any vendor’s product will do the job.
Looking wider afield, the commodity message (certainly for components) is definitely true:
- Hard drives are priced on capacity at the low end (SATA, SMR) and on IOPS at the high end (15K).
- SSDs are priced on endurance & capacity with some variation for performance.
- PCIe SSDs are priced on capacity and performance.
As the exception to prove the rule, NVDIMMs are currently not commodity because there are so few suppliers. This of course will change as other vendors enter the market.
When all products in the market look the same, how do you differentiate and choose between them? Fusion-io had this problem (as well as others) and so looked to provide differentiation by bundling software that made best use of their hardware products (and in some cases were required in order to use them). This strategy will become harder as hardware & software vendors adopt the NVMe standard that means additional drivers aren’t necessary.
Whether one vendor’s products are chosen over another will come down three things:
- Evaluation – give the customers the product and let them try them out. This strategy works in many instances but isn’t scalable.
- Brand – end users will buy brand names based on reliability, value for money and consistency. Building a brand is hard and takes a long time, especially when joining an established market place. Companies like Memblaze will need to produce evidence (white papers, studies, test scenarios) that show how their product fits into the modern data centre, including the current evolution of application delivery. That will mean working with SDS (Software Defined Storage) vendors to build out reference architectures, testing their products against modern databases (and publishing independent results) and working on deployment models such as containers. Memblaze needs to lead, rather than follow, these discussions.
- Partner – The other option is to be the OEM partner of choice with one of the big server or storage vendors. This is a much harder task to achieve unless Memblaze can heavily discount on price to effectively buy the business.
To a certain degree, the storage array market is heading towards commodity. Feature sets are harmonising, performance is more than adequate for most needs (in the all-flash market at least), leaving customers with the agony of choice. When choices are hard to make, customers typically do nothing and the status quo rules. Customer inertia will be one of the biggest issues for Memblaze as the company looks to grow in EMEA and like the whole storage industry will see tough times ahead in 2016.
Comments are always welcome; please read our Comments Policy first. If you have any related links of interest, please feel free to add them as a comment for consideration.
Copyright (c) 2009-2015 – Chris M Evans, first published on http://blog.architecting.it, do not reproduce without permission.