After I hit the “publish” button for yesterday’s SSD article I started thinking of additional things to add:
- I sent a note off to Texas Memory Systems asking them to respond if they’d like. They graciously said they would, so I’m waiting. Clearly though, their product, with multiple FC interfaces, is designed to be a SAN RAM-disk. I hope they look at the Gear6 article as well and respond there too.
- The good folks at Tom’s Hardware tested the Samsung flash SSD and found that it could handle about 2200 IOPS. But that may have been the laptop they were testing it on and not the drive’s in-your-dreams theoretical maximum.
- Another source wrote in to me and said that the cells are actually good for closer to a million read/write cycles. If true, Samsung is silly not to adjust their spec upwards, even to 250k. Engineers can be their own worst enemies sometimes when it comes to promoting a cool new product.
This could be big – if I could just figure out how
I’m still pondering how flash SSDs will play out. Replacing RAM disks – or more likely dramatically extending the market downward – is the low-hanging fruit. Making cheap (relatively) SSDs available will have a lot of folks thinking about their I/O architecture that hadn’t before. A VAR who can figure out the right services that are highly replicable might be able to do well turbocharging certain database and webserver apps.
Comments welcome, of course, as I’m trying to puzzle out what fast cheap SSDs mean for more common computing infrastructures. Moderation on to combat comment spam (10k plus and counting) but registration not required.