Seagate finally made its long-awaited entry into the solid-state disk market this week, counting on its stature as a leading storage vendor and a robust distribution and OEM network to gain traction against rival systems that already are heading into their second generations.
That it took so long for Seagate to make its move would indicate that its first product would be a blockbuster -- one that would outclass the existing field right from the start. But Seagate apparently didn't see it that way. Instead, the company released the Pulsar, a 2.5-inch on-server drive ranging from 50 to 200 GB and sporting 240/200 MBps sequential read/write rates capable of delivering 30,000/25,000 IOPS. Impressive, but hardly mind-blowing.
Company executives claim that it's a true enterprise-class device, rather than a repurposed consumer device, and as such is capable of the rigorous demands of high-intensity environments without failure. That's a nice sentiment, but I would bet that Stec, Intel and other drive manufacturers would put equal enterprise credence behind their systems.
It's clear is that many of the leading storage arrays are starting to move beyond simple SSD performance in favor of their ability to accommodate advanced data-management technologies. EMC, which uses Stec drives for its storage arrays, just came out with the first iteration of the FAST (Fully Automated Storage Tiering) system, which allocates data between SSDs and Fibre Channel or SATA hard drives so the most critical information can be at the ready on the SSD tier. Right now, the system addresses data on the LUN level, although the company is already working out the details for moving down to the sub-LUN/sub-volume level.
Even smaller drive manufactures like SandForce and OCZ are coming out with innovative data-management techniques designed to improve read and write performance. One of them is "garbage collection," according to Enterprise Storage Forum's Herman Mehling. The idea is to eliminate the need to erase entire blocks for each write by identifying only those blocks slated for removal and then pulling them out separately. That capacity can then be used by later writes without having to go through the latency-inducing erase procedure. The technique can deliver improved read/write performance at the outset of the drive's lifecycle and minimize the performance degradation that plagues typical SSDs over time.
The enterprise SSD market is just getting off the ground, so Seagate's late arrival for the pre-show party isn't that big of a deal. With many analysts expecting to see double- or even triple-digit gains in the coming year, there's still plenty of this pie left for sharing.
But the company will have to move fast if it hopes to take a leadership position here. Drive capabilities are the basic requirement. Now the company has to come up with innovative ways to integrate them into existing infrastructures and boost performance beyond the standard.