Topologies for business continuity
- “Bunker site” – Core repository of all production data – Supported by SRM today
- Dedicated sites for prod and DR – Supported by SRM today
- Bi-directional failover – Supported by SRM today
- SRM tech preview – active-active datacenter – zero downtime for planned events; typically limited to metro distance
Typical use cases for active-active data centers: Planned maintenance, disaster avoidance, automated recovery
What do you need for an active-active datacenter solution?
- Stretched storage solution – Distributed data mirroring
- Stretched network solution – VMware NSX or 3rd party solutions; vMotion supports a max of 10ms RTT latency
vSphere Choices for active-active datacenters
- Option 1: Stretched vSphere cluster wit a single vCenter – vSphere Metro stretched cluster (vMSC)
- Option 2: Separate vSphere clusters and a vCenter at each site
Challenges with stretched clusters
- Think about the availability of a single vCenter managing both sites
- DRS and HA are not site aware
- No orchestration or testability today
Active-Active Datacenters with SRM
- Reuse same orchestrated recovery plan for unplanned failures and continuous availability
- Non-disruptive testing
- Local clusters in each site, and DRS/HA work within that site
SRM scenario 1: Local host failure: HA handles failure and no SRM intevention
SRM scenario 2: Disaster avoidance at one site: Execute SRM planned migration with vMotion prior to disaster. gives you an “easy” button
SRM scenario 3: Faster recovery from unplanned failures: Use SRM test capability to prepare for site failure. Execute SRM recovery plan to recover the VMs.
SRM with stretched storage: Existing SRA will continue to work but won’t have active-active support.
The speaker then goes through a full SRM demo, and slides showing the processes happening at each step through a failover and failback.