How people running containers would manage having two different active data centers and not doing anything "legacy" like stretched VLANs? Either running software from both or being able to switch to other DC if other goes down, without losing any traffic.
I'm asking because we mirror two DCs in one city, 2 DCs in another city. Stretched VLANs, synchronously replicated storage. So if DC goes down, vmware just starts the VMs in another and everything is great again. However we can't migrate VMs to another city at the moment.
I've been thinking about having only 2 DCs in total, with BGP or something to route the users to the right place. However I'm not sure how we could replicate the database data or something like that, to be able to do the switchover without losing anything. Colleagues are saying that it's not doable without synchronous storage replication.
We would also have problems when traffic would switch over to another firewall, but I guess that can't be avoided.
I'm hoping to get some new ideas from container world. Our coders are also starting to convert our apps to containers, so would probably need to learn more about this stuff anyway. And having 4 DCs with just few racks in each seems quite wasteful...
Thanks for any ideas!
No comments:
Post a Comment