I've often wanted a larger player.
Fancy spending splashing some cash to buy some flash? Then go to the Lone Star state, where Texas Memory Systems has put itself up for sale. Privately owned TMS has been operating for 33 years. It makes RamSan shared block-access flash memory arrays and has a PCIe flash card line under the RamSan brand as well. It recently …
At least HP most likely won't acquire TMS. Reason: TMS ain't dead enough, they're too cheap and they make good products, which would fill a gap in HP product portfolio. That's a quadruple no-go for HP.
RE: Not HP.
Well, one could point out that 3Par doesn't meet any of the criteria set and yet got bought by hp (before Dell could get their grubby mitts on them). And 3Par seems to be quite a success story for hp.
But I'm not convinced hp will buy TMS. I would have liked them to five or more years ago, but now I see them as more intent on putting flash into storage devices or in the server, not in standalone devices in front of the array like the old RamSan devices.
TMS is El' Crapo
We took a bunch of NAND flash storage vendors in house and tested them all (top 5). I don't want to make this an ad for anyone so I'll refrain from discussing the others. But I will save other people like me precious time. Long story short, TMS has multiple single points of failure and no where close to the performance they promise. Even their "High Availability" box is a joke. I was actually insulted that they even wasted my time. My guess is that TMS wants to sell before their reputation is exposed as junk. Most ALL the other Flash vendors have better products.
Flash is the wave of the future - no doubt there. Look for chip level RAID across entire array, multiple controllers, hot swapable everything, wear leveling and avoiding that pesky Write Cliff Error. Not many have it... but we found it!!!
RE: TMS is El' Crapo
".....Look for chip level RAID across entire array, multiple controllers, hot swapable everything, wear leveling and avoiding that pesky Write Cliff Error....." <Sniff, sniff> Smells like feature sell to me! The "write cliff" is when an SSD is full and write performance drops off as the SSD controller has to search for free space on the "disk" to write to, often by recovering space from deleted items or looking for those free "extents" still available. But that is exactly what happens with ordinary spinning disk too, which is why you'll defragment your drives and run an OS with volume management software that recovers such space. Trying to pretend that Company A's devices don't suffer from the "write cliff" issue is trying to pretend their SSDs work in a different manner to all other SSDs - unlikely. The classis solution is to overprovision - e.g., present a 1GB SSD space as only 800MB so you have a buffer for all that shuffling around of the data and to cope with wear. All solutions I've seen from vendors that claim to avoid the write cliff are simply using software to shuffle data and hide the over-provisioning. Nothing revolutionary or even vaguely unique.
Buy to get ahead not behind
Why would anyone want to buy TMS? Their NAND controller is junk. They have few hardware accelerators compared to other NAND controller vendors, their space/power footprint is significantly large, and they never developed a controller to deal with the challenges of MLC medium (they just use their SLC controller with improved ECC). 2-3 years ago this controller would have been a market leader but the company has not continued to innovate. The execs see the company headed for the crapper and are looking for an exit strategy.