Hi List,
we will soon start investigating into routed/L3-connectivity for Ganeti instances and are wondering, if anyone already has practical experience with such setups. We know there has been work done by Wikimedia[1], which is already well documented (thanks for that!).
Our current apporach assumes that BGP connectivity from inside the instances will be a requirement (so that instances are able to e.g. announce/retract anycast prefixes by themselves). We also do not need live migration (at least not now), so super-fast (network) failovers are not a strict requirement.
The nodes themselves will also run on layer 3-only connectivity with a local bird2 instance, using BGP unnumbered to the upstream switches (see my blog post for further details[2]). This bird2 instance will also be available for downstream BGP sessions (e.g. to Ganeti instances running on the node).
Happy to hear your stories! :-)
Cheers,
Rudi
--
sipgate GmbH - Gladbacher Str. 74 - 40219 Düsseldorf
HRB Düsseldorf 39841 - Geschäftsführer: Thilo Salmon, Tim Mois
Steuernummer: 106/5724/7147, Umsatzsteuer-ID: DE219349391