| *** gyee has quit IRC | 00:02 | |
| *** mikecmpbll has quit IRC | 00:23 | |
| *** mikecmpbll has joined #openstack-swift | 00:24 | |
| *** mvenesio has quit IRC | 00:48 | |
| *** mvenesio has joined #openstack-swift | 00:50 | |
| *** mvenesio has quit IRC | 00:51 | |
| *** itlinux has quit IRC | 01:04 | |
| *** armaan has quit IRC | 01:23 | |
| *** armaan has joined #openstack-swift | 01:23 | |
| *** itlinux has joined #openstack-swift | 01:31 | |
| *** klamath has quit IRC | 02:37 | |
| *** klamath has joined #openstack-swift | 03:46 | |
| *** klamath has quit IRC | 04:00 | |
| *** germs has quit IRC | 04:17 | |
| *** germs has joined #openstack-swift | 04:17 | |
| *** SkyRocknRoll has joined #openstack-swift | 04:29 | |
| *** germs has quit IRC | 04:40 | |
| *** mikecmpbll has quit IRC | 04:49 | |
| *** germs has joined #openstack-swift | 04:56 | |
| *** germs has quit IRC | 04:56 | |
| *** germs has joined #openstack-swift | 04:56 | |
| *** germs has quit IRC | 05:01 | |
| *** SkyRocknRoll has quit IRC | 05:03 | |
| *** klamath has joined #openstack-swift | 05:57 | |
| *** klamath has quit IRC | 06:02 | |
| *** germs has joined #openstack-swift | 06:18 | |
| *** SkyRocknRoll has joined #openstack-swift | 06:19 | |
| *** zaitcev has quit IRC | 06:23 | |
| *** germs has quit IRC | 06:24 | |
| *** SkyRocknRoll has quit IRC | 06:33 | |
| *** armaan has quit IRC | 06:53 | |
| *** armaan has joined #openstack-swift | 06:54 | |
| *** pcaruana has joined #openstack-swift | 07:01 | |
| *** SkyRocknRoll has joined #openstack-swift | 07:28 | |
| *** armaan has quit IRC | 07:43 | |
| *** armaan has joined #openstack-swift | 07:43 | |
| *** klamath has joined #openstack-swift | 07:58 | |
| *** klamath has quit IRC | 08:03 | |
| *** germs has joined #openstack-swift | 08:20 | |
| *** germs has joined #openstack-swift | 08:20 | |
| *** germs has quit IRC | 08:25 | |
| *** gkadam has joined #openstack-swift | 08:43 | |
| *** mikecmpbll has joined #openstack-swift | 08:52 | |
| *** onovy has quit IRC | 09:44 | |
| *** cshastri_ has joined #openstack-swift | 09:58 | |
| *** klamath has joined #openstack-swift | 09:59 | |
| *** klamath has quit IRC | 10:04 | |
| *** cshastri_ has quit IRC | 10:10 | |
| *** germs has joined #openstack-swift | 10:21 | |
| *** germs has quit IRC | 10:21 | |
| *** germs has joined #openstack-swift | 10:21 | |
| *** germs has quit IRC | 10:25 | |
| *** geaaru_ has joined #openstack-swift | 10:26 | |
| *** geaaru has quit IRC | 10:28 | |
| *** gyankum has joined #openstack-swift | 10:34 | |
| *** gyan_ has joined #openstack-swift | 10:34 | |
| *** gyan_ has quit IRC | 10:40 | |
| *** gyankum has quit IRC | 10:40 | |
| *** onovy has joined #openstack-swift | 10:44 | |
| *** armaan_ has joined #openstack-swift | 11:54 | |
| *** armaan has quit IRC | 11:54 | |
| *** klamath has joined #openstack-swift | 12:00 | |
| *** gyankum has joined #openstack-swift | 12:01 | |
| *** gyan_ has joined #openstack-swift | 12:01 | |
| *** klamath has quit IRC | 12:04 | |
| *** armaan_ has quit IRC | 12:16 | |
| *** armaan has joined #openstack-swift | 12:17 | |
| *** torgomatic has quit IRC | 12:27 | |
| *** gyan_ has quit IRC | 12:28 | |
| *** gyankum has quit IRC | 12:28 | |
| *** torgomatic has joined #openstack-swift | 12:30 | |
| *** ChanServ sets mode: +v torgomatic | 12:30 | |
| *** gyankum has joined #openstack-swift | 12:30 | |
| *** gyan_ has joined #openstack-swift | 12:31 | |
| *** armaan has quit IRC | 12:37 | |
| *** armaan has joined #openstack-swift | 12:46 | |
| *** armaan has quit IRC | 12:50 | |
| *** vinsh has quit IRC | 12:56 | |
| DHE | a (minor) bandwidth issue for multi-region situations occurs to me. I'm looking at building a multi-region (like, 4) cluster with 1 or 2 replicas per region. that means that PUTs are going to write a lot of copies out around the WAN links... | 13:13 |
|---|---|---|
| *** persia has joined #openstack-swift | 13:15 | |
| *** Renich has quit IRC | 13:25 | |
| *** gyan_ has quit IRC | 13:38 | |
| *** gyankum has quit IRC | 13:39 | |
| *** Renich has joined #openstack-swift | 13:46 | |
| *** cshastri_ has joined #openstack-swift | 13:48 | |
| *** cshastri_ has quit IRC | 13:59 | |
| *** klamath has joined #openstack-swift | 14:01 | |
| *** klamath has quit IRC | 14:06 | |
| *** klamath has joined #openstack-swift | 14:07 | |
| *** jistr has quit IRC | 14:09 | |
| *** jistr has joined #openstack-swift | 14:14 | |
| *** jistr has quit IRC | 14:22 | |
| *** ianychoi_ has joined #openstack-swift | 14:22 | |
| *** ianychoi has quit IRC | 14:25 | |
| *** jistr has joined #openstack-swift | 14:28 | |
| *** armaan has joined #openstack-swift | 14:42 | |
| notmyname | DHE: yes | 14:51 |
| notmyname | DHE: in a global cluster, I'd recommend setting up per-region memcache pools. you don't want a proxy in region A asking a memcache server in region B for cached data | 14:52 |
| DHE | well, yes that makes sense. | 15:02 |
| DHE | my main concern is I'm looking at 1.5 to 2 gigabits/second of raw uploads | 15:03 |
| *** hseipp has joined #openstack-swift | 15:06 | |
| notmyname | have you come across write affinity yet? | 15:09 |
| notmyname | disclaimer: I hate write affinity. it's almost never what you actually want | 15:09 |
| notmyname | write affinity let's you defer inter-region communication until later (ie not during the client request) | 15:10 |
| notmyname | it will make your utilization lumpier and it intentionally puts a ton of data in handoff locations (ie lots more work for the replication process to do later) | 15:10 |
| notmyname | but, if you have bursty/seasonal workloads and if your ingest is lower than your ability do do the deferred work and if your users cannot tolerate additional latency on writes, then maybe just maybe write affinity is a thing that could help | 15:11 |
| *** hseipp has quit IRC | 15:12 | |
| DHE | yes I've read all the online docs (besides python API level stuff) | 15:19 |
| DHE | no, it's very continuous and borderline constant bitrate | 15:19 |
| notmyname | ah | 15:20 |
| notmyname | unlikely that write affinity would make your job easier, then | 15:20 |
| DHE | what I do have in mind is a strange idea. there's a hundred-ish uploads of smallish (4 megabytes or less typically) files per second. it might be tactically useful to upload to a proxy in a different region so the multiplication factor of distributing to object servers is also distributed... | 15:22 |
| DHE | does that make sense? | 15:22 |
| DHE | these 4 sites are in a ring | 15:22 |
| DHE | even though all the uploads would originate from the same region | 15:23 |
| *** armaan has quit IRC | 15:26 | |
| *** szaher_ has quit IRC | 15:41 | |
| *** szaher has joined #openstack-swift | 15:54 | |
| *** gkadam has quit IRC | 17:02 | |
| *** SkyRocknRoll has quit IRC | 17:43 | |
| *** ukaynar_ has joined #openstack-swift | 18:28 | |
| *** bkopilov_ has quit IRC | 18:37 | |
| *** bkopilov has quit IRC | 18:38 | |
| *** gkadam has joined #openstack-swift | 18:41 | |
| *** bkopilov has joined #openstack-swift | 18:49 | |
| *** bkopilov_ has joined #openstack-swift | 18:51 | |
| *** geaaru__ has joined #openstack-swift | 19:27 | |
| *** geaaru_ has quit IRC | 19:28 | |
| *** bkopilov has quit IRC | 19:38 | |
| *** bkopilov_ has quit IRC | 19:38 | |
| *** itlinux has quit IRC | 20:07 | |
| *** ukaynar_ has quit IRC | 20:08 | |
| *** ukaynar_ has joined #openstack-swift | 20:08 | |
| *** ukaynar_ has quit IRC | 20:12 | |
| *** itlinux has joined #openstack-swift | 20:46 | |
| *** ukaynar_ has joined #openstack-swift | 21:54 | |
| *** pcaruana has quit IRC | 22:24 | |
| *** ukaynar_ has quit IRC | 23:20 | |
| *** ukaynar_ has joined #openstack-swift | 23:20 | |
| *** ukaynar_ has quit IRC | 23:23 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!