Competition implies adoption which equals price drops
This I like.
Startup Pliant is launching its Lightning solid state drive (SSD) today, with SAS interfaces only and STEC-beating performance claims. A fair amount of Pliant's marketing literature speaks of SSDs being much faster than hard disk drives and compares an all-HDD configuration with a hybrid SSD/HDD configuration to satisfy a TPC- …
A good summary Chris. It is good that STEC have potential competition. This will help drive the SSD (Enterprise Flash Drive) market. As you say, Pliant need to convince the likes of EMC, IBM etc to use the their drive. One problem is that the EMC Symmetrix has a FC backend (like over enterprise storage arrays), not SAS, which means that the drives would require a FC to SAS converter.
Also as I have blogged (www.spacrc.typepad.com), I don't believe it is all about max IOPS numbers. I don't believe most customers have a need for the max IOPS performance of these SSD's. Most customers would perhaps only currently drive a few thousands IOPS to each drive, and the current generation of enterprise storage arrays cannot exploit the max IOPS potential of these SSD's. There is far more to it than the max IOPS.
Anyway, I will watch this SSD battle with much interest.
Pliant (via the CEO's blog) has been getting lots of people excited for a year now about the number of IOPS it would do in ==mixed read-write== workloads.
This is hugely important, because both Intel and STEC performance drops to around 10% of the advertised numbers with mixed read/write workloads, making these devices nearly useless (from an ROI perspective) for transactional database applications.
So....when I looked at the Pliant spec sheet I expected to see some of the numbers they've been bragging about...but they aren't there. In fact there is NOTHING there about the test workload.
Nothing but the same useless 100% read-only IOPS, and Pliant doesn't even bother to disclose the size of the IO requests.
This is FAR WORSE marketing behavior than Intel and STEC, not better, and a mistake on Pliant's part...IMO. Unless they are hiding something...
If Pliant had really achieved the breakthrough they have been claiming, all it would have taken was 30,000 read/write IOPS to beat both STEC and Intel and STEC by 5x-10x. Even a measley 15,000 would beat STEC by 3x, and put them way beyond Fusion-IO to boot.
Hey Pliant...Real-World = Read/Write!!! How about getting behind your product with some real numbers? A 60/40 mix of 8K IOs would be just fine, thanks. Oh...and don't forget to specify the queue depth.
Or better yet, rather than just talking about TPC-C, why not publish one? Or TPC-E?
Pliant spec sheet says the 2.5" SSD does 120,000 IOPS, but a post by the CEO on the Pliant blog from July shows they need four of them to get to 50,000 IOPS:
What's the deal???
Biting the hand that feeds IT © 1998–2021