Adaptec's caching card uses MS code
MaxIQ uses Redmond algorithms
Adaptec has announced benchmark results for its MaxIS solid state drive (SSD) caching card and revealed it's based on licensed Microsoft code.
The MaxIQ card is equipped with 32GB Intel X25-E SSDs and is an add-on card to Adaptec's RAID controllers such as the 5Z, which link to SATA and/or SAS hard drives.
In effect Adaptec has made a PCIe-connected flash memory card that happens to be a RAID controller as well. As such it competes with pure SSD PCIe cards that have no RAID functions on them, such as Fusion-io's ioDrive and LSI's SSS6200.
Adaptec claims its MaxIQ card is more cost-effective than other PCIe flash cards.
The benchmark simulated a brokerage application database environment with 50,000 users, and showed a 3.2X increase in system I/O (input/output) performance over the best results of a similar system without SSD caching. Adaptec claims such a performance increase "can translate into a savings in capital and operating costs of up to 70 per cent".
Adaptec said it licensed ideas from Microsoft about using SSDs as read caching devices. Its release about the benchmark quoted David Kaefer, general manager of Microsoft's Intellectual Property Licensing, saying: "When our datacenter team came up with some innovative ideas around using solid state devices as read caching devices, we determined it made good sense to license these advances to Adaptec because Microsoft itself doesn't sell these types of products."
Adaptec is up for sale, following activist investor Steel Partners' takeover of the board. What we appear to have here is a demonstration of Adaptec's product prowess and intellectual property assets - potential buyers take note. ®
Re: Re: How long will they last
"It may be unusual to have a computer component with a finite lifetime......"
Those MTBF figures quoted on just about abso-bloody-lutely everything. What are those for then?
Everything has a finite life. Some things just have one that's so long that obsolescence tends to preceed failure.
Re: How long will they last
I thought that modern SSDs had several years of life in them even under fairly intensive usage. Also, this particular mode of failure doesn't exactly creep up behind you. It isn't too hard to actually record the lifetime usage and let the user decide when it is time to replace the device. Said replacement will probably only be a few pounds by then anyway. These are caches, so they might be fairly small in absolute terms.
It may be unusual to have a computer component with a finite lifetime, but it isn't unmanageable.
Ehh, the X25-E is SLC flash, which is pretty decent about lasting for a while. Add to that the fact that the drive has something like 25% more physical flash onboard than they let you access, you get lots of room to replace failed sectors. I imagine a drive like that would last quite a while.
Really though the benefits from caching come from data that's got a nice high read-write ratio, and since you can read an SSD as many times as you'd like, that translates to a nice slow wear-out.
Though that drive's performance is not even kind of a little bit similar to the Fusion-IO, so I suppose it depends on what you need, speed-wise.