In case you missed it, in December, we ran a webinar on a topic that we’ve been asked about frequently by end-users: Can you run SvSAN as a 3-node deployment?
Mark Christie, Principal Technical Architect at StorMagic started the webinar by briefly outlining how SvSAN can be deployed as a 3-node system. With our approach to 3-node clustering, if a node fails or is taken offline the remaining two servers still maintain high availability. This isn’t common in the industry as most other HCI solutions require at least 4 nodes to achieve the same results.
Mark then takes us through a live demo , where he simulates a server failure. . He explains how the deployment remains highly available during the downtime, through continued synchronous mirroring between the remaining nodes. He also explains how the witness service is able to keep track of the situation and ensures smooth operations when the server comes back online.
Q&A questions answered in the webinar include:
- You’ve demonstrated a host failure in this setup, what would happen if I want to carry out maintenance on these systems?
- Would you recommend using the same witness [as used for 2-node deployment] for these environments?
- Can I do this with SvSAN’s other features, like encryption?
- How do you prevent split brain clusters in case of a failing node coming back?
- In the event of failure or maintenance, is manual intervention absolutely necessary?