[10:31] Morning [10:59] Morning. [13:03] Morning peoples [13:16] Hi. [13:41] o/ [16:55] morning, peeps [17:28] ChinnoDog: You like to nom on peeps ? [17:52] Sometimes. Especially when they are roasted over an open fire until the sugar carmelizes. [18:23] pleia2 eats them that way [18:32] ChinnoDog: invite me next time you build a fire and have ample stock. [18:33] Sounds yummy [19:01] only on special occasions [19:05] There should be more of those. We could have a peep roasting contest. [19:15] I don't actually like them :) [19:15] but on a freezing bethlehem night, it's an acceptable thing to do [19:20] They are good at negative degree temperatures. (I don't like them either.) [19:41] waltman: yeah, going to Ko's guys really was the solution! [19:41] major help === calvin_ is now known as MidgetTurkey [19:43] Anyone have advice for managing 124tb of data while maintaning backups and redundancy? [19:43] Midget? Must be wild. Domesticated turkeys can get up to 70lbs. [19:43] heh [19:43] wait what... I'm already logged in a nother window [19:43] sigh [19:44] silly me [19:44] haha [19:46] question still stands [19:46] Anyone have advice for managing 124tb of data while maintaning backups and [19:46] redundancy? [19:46] whoops sorry [19:50] 124tb is a lot [19:51] MutantTurkey: I'd just print a copy out, personally. [19:51] just print it in hex, and you can type it back in if something goes wrong. [19:56] MutantTurkey: What type of uptime do you need to mantain? [20:06] heh [20:06] ChinnoDog: as up as possible [20:06] i need more like, equipment recommendations [20:06] what kind of server do i buy that has that ind of capacity [20:15] You are going to have to define these parameters before you can answer your question [20:15] For example, on my home computer I can tolerate a few days of downtime because of a disaster. If I lose my files no one is going to die or fortunes lost. [20:16] I only use RAID 0. I run external backups to a slow external USB drive and also to the internet. [20:16] If a drive dies I will restore from external USB. If the apartment burns down I will order a backup of my data from Crashplan. [20:18] My server will not stay up if a drive fails. In order to do that I would need more disks and have to spend more money and I would /still/ need an external backup plan [20:20] At 124tb the cost of the redundancy is cheaper depending on how resistant to failure you want to be. Theoretically you could build a 124tb volume using 42 3tb drives + 1 for RAID5 and be resistant to the failure of one disk [20:20] You can figure out how frequently you will be replacing disks using the MTBF ratings of the drives [20:20] You can also calculate the likelyhood of two drives failing at the same time [20:29] downtime, one day. [20:29] we have a raid6 array right now [20:29] I haven't seen it yet so I don't know what's there physically [20:37] That sounds suitable for that number of drives so long as you don't waste too many on parity. [20:42] yeah? [20:46] Using RAID 6 you are utilizing the capacity of all the drives minutes the 1+ drives for parity. You can scale the number of drives for parity based on your MTBF calculation. [20:46] s/minutes/minus/ [20:50] Why hasn't anyone written a RAID accelerator using OpenCL? is it just me or is that a natural application? Video chipsets with stream processors are creeping into server boards, might as well make use of them. [20:52] For that matter OpenCL would be perfect for file system compression too. [21:43] MutantTurkey: cool [21:43] so they actually knew about a library? [21:59] well no, but they are python pros [21:59] so we did it in python [22:00] turns out that function only has one implementation [22:00] the origional one [22:06] MutantTurkey: hah [22:36] setting up nagios now