Regarding Violin Systems and its race for All-Flash Array glory

Violin Systems has launched its latest VXS 8 all-flash array into what it calls the extreme performance all-flash array market.

Its a hot box with latency down to 50μs when using NVMe over Fibre Chanel, and dedupe and compression are switched off. The VXS 8 has cloud-delivered predictive analytics, with an augmented reality smart phone app as an end-point for this, and it has a coming move away from its costly proprietary flash drives, the Violin Inline Memory Modules (VIMMS), towards using TLC (3bits/cell) 3D NAND SSDs connected to the VIMM carrying-board.

The marketing pitch is extreme and consistent low latency, high-IOPS performance equivalent to the NVMe-oF startups (E8, Excelero, Vexata), that’s claimed to be 5x to 7x faster than traditional AFA vendors (Dell EMC, HPE, IBM, NetApp, Pure) while offering the same levels of enterprise data services as these mainstream trad AFA players, and which the NVMe-oF startups lack.

This is Violin’s self-declared market niche. Can it succeed, having failed before?

Violin says it has more than 100 Fortune 500-size customers running mission-critical applications. This is the base on which it must build. It says the three most important AFA attributes to them are low latency, consistent performance  and enterprise data services. Their applications are, Violin says, tier zero ones, which includes OLPT, DBMS and VDI, and they offer real-time insights to help the customer’s operations.

Upcoming applications are in the AI, IoT, trading and supply chain areas.

It says its customers can make greater profits from decreased transaction times, an accelerated supply chain and faster insights into such things as financial trading calculations. All this depends upon the aforesaid consistent low latency and high-performance.

If it can demonstrate that, and show that its customers need less IT infrastructure to run their their tier 0 apps; meaning fewer servers and therefore systems and application SW licenses, fewer network links, less cooling and so forth, then it has a chance.

Violin says using its gear means that customers have a lower IT total cost of ownership with savings so great that the Violin storage can almost come for free. 

It claims that mainstream AFA vendors; Dell EMC, HPE, IBM, NetApp, Pure along with DDN Tintri and Nutanix, are only suitable for tier 1 applications and mixed workload storage, not tier 0 apps. 

Violin’s view of its market landscape

This is Violin’s claimed market niche; tier 0 application performance for F500 enterprises that must have enterprise data services.

But the mainstream AFA suppliers all have NVMe-oF technology speed ups coming and the hot box startups will get the data services Violin says they need.

How long does Violin have before its window of opportunity closes?

It has to stay at the top of the performance (IOPS, low latency) ranking. It has a roadmap to migrate away from proprietary VIMMs towards using 3D NAND TLC (3bits/cell) SSDs, with microcode additions giving it direct control over the flash chips inside the SSDs. That should help counter the inherent slowness of TLC NAND compared to MLC NAND, and aid in reducing its costs and so helping fund product and business infrastructure development.

In our view the next two years are crucial. Violin has made the necessary first product refresh steps. It now has to demonstrate it’s a stayer, with the installed base taking up the new kit and its channel finding new customers for it.

Performance-wise it has to stay at the top of the tree, and if things like Optane caching in accessing servers are needed to do so then it must be done.

Its crucial, in our view, that Violin can believably claim a performance crown and retain it.  Fail at that and its whole marketing edifice crumbles.

The Soros Fund, Violin’s owner, faces a dilemma if the company falls behind in the performance stakes. It would have to raise it stakes to develop faster product or quit. We should see which way the wind is blowing by the end of 2020.