An SSD For The Rest Of Us
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.
on October 21st, 2006 at 11:18 pm
The killer application for SSD is probably in mobile devices. That seems to be the driving force for SSDs, at least so far. HDDs and mobile computing (is there other type now?) do not mix. Have you dropped your 30GB iPod on the floor recently, or used your ‘working horse’ notebook on the table in moving train/plane? It is not funny!
There is a business case for SSD - it will be everywhere, GPSs, cell phones, players, TVs with DVR, web and file servers, notebooks, toasters… did I miss anything?
If I was a HDD manufacturer I would be seriously worried. The only thing that can keep them alive is the lower price for HDDs compared to SSDs, but not for long.
I really hope Samsung will drop the prices for the SSDs real soon now, after all, there is market waiting for it! Although recently they pleaded guilty on RAM prices fixing…so… I don’t know.
There are still some issues with SSD though, no OS knows how to handle them properly, and there are just a few flash optimized file systems (for Linux) - more like proof of the concept, but they will catch up.
ZFS looks like it will be able to work with SSD just fine. Micro$soft’s Vista will try to boot ‘faster’ from hybrid flash drive cache - only to boot, probably because NTFS is just a thing from the past (I loved your article - ‘Bring me the head of WinFS’s project’
I read somewhere about hypothetic Google OS, it is suppose to be optimized to work exclusively with flash, and boot from USB flash drives. You plug in the Google USB flash, boot, and there you go, all your email, calendar, office tools - online, on demand. No licensing fees, no ‘genuine advantage’… infect me with all the viruses you want - I will reimage the flash.
What I really liked is Tom’s Hardware charts… web server’s benchmark is just sweet! Given the nature of web servers - read only files - a flash drive will probably last ‘forever’. There is a limit on r/w cycles for flash, but the limit for just read cycles should be much bigger, right?
Not much corp. web sites are bigger than 8GB, it makes me wonder why am I still using HDDs for web servers.
What I really hope is one day SSDs to get into the file servers! That will save so much space and power. Sun’s Thumper with its ~70kg will look like a dinosaur from the Jurassic period of the storage. “- Look son, we used to use these for storage… -geee dad, how did you survive as an IT guy in that era?”