Why gigabyte should revive the I-RAM concept
Posted by xipietotec on October 10, 2007
Gigabyte created a neat little niche product called the I-RAM, its basically a board that allows you to plug in up to 4gb’s of DDR ram and run that through a Sata 1 connector, giving you essentially a massive ramdisk. It requires no drivers, and is blazingly fast. Unfortunately in that incarnation it was doomed to failure. Why?
- The device itself is expensive, $120 is about the cheapest you’ll find it online.
- DDR ram is very expensive. Cheapest 1gb sticks you’ll find online cost about $50.
- Sata I is slow.
Now, around 2006 they showed off the I-RAM 2, it was everything that the I-RAM One wasn’t, and they bloody shelved it. It used Sata II (3.0gbs/s), DDR2 (1gb sticks can be had for $25 each), and supported up to 8gb’s of ram.
Now, why might this be useful? Oh lots of applications. Say you’re an independent web developer, you could easily use an 8gb solid state swap space for handling data or streaming media. Next, your HTPC, streaming large portions of the video through that swap. Putting your /var /tmp and /swap on it for large NAS’s. There are many many lovely uses for it, and Gigabyte shot themselves in the foot on the first go around.
One of the biggest uses I could see, is use in a “poor-man’s data-center” Large solid-state storage and server motherboards with lots of ram are bloody expensive and require ECC ram. A Cheap off the shelf unbuffered solution that totals less than $500 is a bloody steal in comparison. (That’s say $100-150 for the device, and another $200 for 8gb’s of ram, at my $25/gb price, for a total of $300-350). Gigabyte simply needs to actually market it as such.
I can think of one huge potential customer, who makes most of their own servers using cheap off the shelf parts: Google.
And one smaller customer who would love to put it to work in his home NAS: Me.
