IRC Logs for #crimsonfu Friday, 2018-10-26

*** commavir_ has quit IRC00:25
*** pdurbin has joined #crimsonfu00:29
*** commavir has joined #crimsonfu00:34
larskspdurbin: I dunno; I'm actuall talking for 10 minutes, but I haven't paid much attention to what else is going on.01:09
*** commavir_ has joined #crimsonfu05:51
*** commavir has quit IRC05:55
*** jri has joined #crimsonfu07:29
pdurbinlarsks: ok, thanks. :)10:40
dotplusafter brief reading, it's not clear whether MOC is supposed to be _an_ instance of OCX or _the_ (singleton) implementation of OCX or a participant in a larger thing which is OCX. Also, is it determinedly just openstack or is that just "one" cloud provider?13:30
dotpluswhatever the answers, it sounds like you have interesting things going on up there. Perhaps next time I go to Manchester, I'll try to stay in Boston instead of just flying through BOS (because flying to BOS and getting a car from BOS->MHT is more reliable than trying to fly PIT-MHT)13:32
pdurbinWhat's OCX?13:32
larsksdotplus: I'm not sure I have good answers to those questions. They sound like good ones to ask next week, though :)13:33
larskspdurbin: Open Cloud Exchange.13:33
dotpluspdurbin: open cloud eXchange, tsee the page you linked13:33
larsksOne of their goals is that someone using an openstack environment can ultimately make sure of services/resources from *another* openstack environment.  E.g., you get your compute from provider A but your storage from provider B.13:34
dotplusI'll be in Nashville next week: https://www.usenix.org/conference/lisa1813:34
larsksThat doesn't work right now, but they are working on the necessary changes (e.g., keystone-to-keystone federation)13:34
dotplusouch. last time I remember (admittedly, I haven't touched openstack for nearly 2 years since I left Cisco) they were working on federation then.13:36
dotplusIt's hard to believe that storage from a different vendor than your compute provider would be acceptable from a performance point of view for anything but archive/backup purposes; wouldn't it latency be an issue?13:39
dotplusI dunno, perhaps the distinction would be: block storage should be close, but object storage could be from elsewhere13:39
dotplusif I had to do it now, I'd do it clientside use terraform, such as: OCI provider for compute instances, block storage; AWS provider for lambda. And indeed, I don't think there's a reason why you couldn't have terraform say: "use OpenstackA for this set of state/modules, use OpenstackB for this other set"13:45
dotplusI guess the really hard parts of all this are 1) networking 2) IAM13:46
*** jri has quit IRC15:45
*** chasmo77 has quit IRC15:48
*** chasmo77 has joined #crimsonfu15:50

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!