A couple of weeks ago Dell EMC announced the demise of the once promising DSSD all flash array. They are planning to incorporate DSSD technology into their other products.
As StorageMojo noted 4 years ago, DSSD developed a lot of great technology. But for whatever reason – perhaps turmoil associated with Dell’s purchase of EMC? – EMC’s less-legendary-than-they-used-to-be salesforce couldn’t move the boxes.
The StorageMojo take
The AFA market is full up with aggressive competitors, and any new AFA is going to face tough sledding. The entire AFA market is facing a dry spell as flash prices have firmed up by 25% in the last six months, and look to remain high well into next year. Hybrid arrays will benefit, as disk capacity has resumed its upward march thanks mostly to shingled magnetic recording, and drive prices are still dropping.
I suspect DSSD’s Silicon Valley ethos had problems integrating into EMC, a company known for full contact politics. And the purchase agreement undoubtedly had multiple milestones, which over time became a straight jacket instead of a carrot.
But the plan – even if it’s just a face saving gambit – to incorporate DSSD technology into other products, has StorageMojo’s full support. As StorageMojo stated 3 years ago:
. . . EMC could use DSSD as a VMAX backend, probably with thrilling performance. So why not mention that? You get all the wonderful software features of VMAX – and a big performance boost!
That was the obvious play then. Today, even a massive shot of hardware Viagra – such as DSSD – can’t save VMAX. If only they’d listened.
Courteous comments welcome, of course.
Bolting things together that were never meant to be together are EMC’s greatest strength (and weakness) I’m sure they’ll find a way to hack the DSSD kit into a VMAX and call it something special.
There were several reasons why DSSD didn’t sell well:
Foremost, customers didn’t understand how to consume it over PCIe; a direct attached PCI express fabric mesh was completely foreign to most shops.
The price. Even at extreme discounts it was very expensive to own.
Data services. It was designed not to have them, yet customers wanted inline compression, encryption, and snapshotting capabilities.
For 99% of customers, the much slower in server flash was good enough.
There was an amazing amount of design genius wrapped up in the product between the cubic raid, elimination of controller ASICs in the IO path, and power regulation to achieve the highest transfer rates and flash endurance rates in the industry.
Unfortunately, very few had a use for it…