Messages in this thread | | | Subject | Re: Building Big Ass Linux Machine, what are the limits? | Date | Mon, 28 Sep 1998 10:06:20 -0400 | From | Theo Van Dinter <> |
| |
| Where you want safe delayed write, than controllers with cache | can have one important advantage: battery backup for the cache memory | on board; | Whether the cost is justified - this depends on application. | | Assuming the server will never die for any reason (including RAM failure, | UPS failure, etc) - extra RAM is surely the way to go, but how can one | be sure?
I read a paper (http://www.netapp.com/technology/level3/3002.html) the other day about the WAFL filesystem (used on the Network Appliances at www.netapp.com). The NetApp boxes use NVRAM to record a log of NFS activity since the last write to disk, and it does a consistency "snapshot" at least once every 10 seconds. Basically it comes down to this: The filesystem on disk is always consistent (ie: no fsck needed after reboot), and the nvram stores what changed since the last write (battery backed up so a power failure doesn't kill it) ...
It's pretty nifty, in the middle of it doing things, you can just flip the power switch. Turn it back on, wait about a minute for the thing to boot, and you're back up and running.
I'm not sure if the design is "free for use", but it would be very nice to have as an addition to Linux if it is available for general use.
-- Randomly Generated Tagline: "Besides, I think [Slackware] sounds better than 'Microsoft,' don't you?" - Patrick Volkerding
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |