Openstack what I wish I knew
Below are some points of things that I wish I knew before jumping into Openstack. It wouldn't have changed the plan dramatically but would have made life easier. (I will update this as we go further along this journey.)
- Openstack has had dramatic changes over it's versions. This means documentation can be fragmented or outdated. The big one in our instance was that we initially set out to use Gluster as the storage backend as we had experience using it. Finding documentation on configuring openstack to use is redialy available on the internet. I built puppet profiles to dynamically build and expand gluster to account for the expectation that the initial cluster of 8 servers will blow out to 16 withing 12 months and wanted to make expanding easy (We were using a "hyperconverged" setup where local storage on each blade was part of shared storage. We run VMware VSAN in our primary production datacentre so this would be similar). It was at the point of having this completed and building openstack that it started reporting that the gluster storage module is unknown and further rearch showed gluster was depreciated. We ended up moving to Ceph.
- Openstack isn't a replacement for VMWare. If you are looking at replacing VMWare with Openstack they aren't a competitor to each other. For enterprise environments I would definalty keep it on VMware (or Hyper-v if you are that way enclined), we were moving our non-production environments for a Saas product. Things like being unable to attach ISO without using some workarounds is painful. Think Openstack more like AWS EC2 but you manage the infrastructure.
- Expect to have to go into the database at times. There have been a few times I have had to go into the database and update records. Something I detest but either records were corrupt or there is no way to set a value through the cli or web interface.