Hi, your questions are quite complex, I don't know if anyone can give you defintive answers since there are so many requirements involved and careful planning is key. Maybe you could start with [1] to get an overview how something like that can be done. One scenario is basically to have two ceph clusters and replicate the RBDs from primary A to secondary site B. In case site A goes down you can promote the rbd images in B to be primary and your openstack should continue to work. The video also covers multiple openstack clusters which have the same data (and also an rbd mirror between the two ceph clusters), check it out, maybe a couple of your questions are covered. Regards, Eugen [1] https://www.openstack.org/videos/summits/austin-2016/protecting-the-galaxy-m... Zitat von KK CHN <kkchn.in@gmail.com>:

few more points to clarify... 1. The VMs in a DataCentre must auto migrated to the DR setup. ( Also the updates / writes to VMs in production in DC should be reflected to the DR copies of VMs like incremental backups. How to achieve this / or what needs to be employed to achieve this requirement.

( What all are the S/W and H/W requirements to achieve the above setup if both DC and DR is planned use OpenStack latest version/s Wallaby/ Xena/ yoga ? )

2. In the above setups once the DC down or stopped for maintenance, How the IP addresses of each VMs is managed automatically to make all the (application/database server) VMs up and running in DR in case DC down . Is this can be automated? how?

Eg: When a VM say X is running in DC it may have an IP 10.10.0. X when it is replicated in DR then it will be with the same IP address right (10.0.0.2) ? But DR Network may be different and cannot have the same IP address as DC right ? Do we need to manually set an IP (Say 10.20.0.X )for each VM which is going to run from the DR site ? Then what about the firewall rules in DR do we need to manipulate for each VM for making the DR up ? Is there a way to automate this ?

OR what the automatic mechanism to handle this IP setting up issue ? How normally folks manage this scenario ?

Also once your DC site recovered, then We need to Fail back to the DC site from DR with all changes happened to the VMs in DR must be reflected back to the DC site and Fail back.. .How to achieve this ?

Kindly shed some light on this with your experience and expertise.

What to do ? Where to start ? Which approach to follow to set up a Best failover DC to DR and Failback solution.

Thank you, Krish

On Tue, Oct 11, 2022 at 5:38 PM KK CHN <kkchn.in@gmail.com> wrote:

...

List,

We are having a client DC running on HP( HP simplivity) HCI servers, With VMware ( Vsphere 7.0) only few VMs running on it. (11 VMs maximum all Linux VMs).

The DR site also having the same HCI setup another location. ( The VMs are replicated to DR site with HP simplivity).

We are planning to use Openstack for both DC and DR solutions with Wallaby or Xena version with KVM as hypervisor to replace the proprietary S/W and H/W vendor locking.

The requirement is to Setup a Stable DC- DR solution.

Totally confused How to setup a best Dc- DR solution for this purpose.

The DR setup can be possible / advisable with Zero down time ?( or manual DR site uping with downtime of hours ) ?

What are the available/suggested DC-DR replication mechanisms for high degree of application data protection and service availability?

Kindly advise..

Thanks in advance, Krish