Stretched Cluster is something different as stretched cluster adds an another dimension to the traditional HA cluster – a geographical location.
In normal vmware HA the failure protection is assured with a traditional HA and you are able to maintain your hardware, hosts by motioning your VMs to other hosts, during the maintenance tasks. However, when a whole site fails, you’re not protected.
You may have a backup and replication product and replicating your workloads elsewhere so your VMs are in “standby” state and just needs to be powered On. This is a good way to start, but the ultimate protection is definitely VMware Stretched cluster scenario.
The main problem of stretched cluster scenarios was often a shared storage which needed to have a hardware support for synchronous replication between the main site and remote site. Note synchronous replication is when a disk IO is performed by the application or by the file system on the primary site, it waits for the IO acknowledgment from the local disk and from the secondary server, before sending the IO acknowledgment to the application or to the file system.
With VMware vSAN, this hardware requirement for synchronous replication is no longer necessary as there is no any SAN hardware required. VMware vSAN is using local disks within each host to build this shared storage, the vSAN. And for the stretched cluster scenario it is VMware vSAN which “stretches” the storage between the main site and remote site without using SAN or NAS hardware. Both sites are active so you can your VMs on both sides simultaneously. (The other site is not just waiting for the passive or standby VMs to be powered ON).
VMware Stretched Cluster Requirements:
Whether you’re using traditional SAN device on both ends or VMware vSAN, you have hard requirements – network. The network latency and throughput have to meet the system requirements so VMware can certify the solution.
(Note that there are other hardware requirements for implementing VMware vSAN).
An example of network requirements for VMware vSAN:
Network Requirements between active – active sites
- 10 Gbps connectivity or greater
- < 5-millisecond latency RTT
- Layer 2 or Layer 3 network connectivity
Network Requirements from active – active site to witness site
- 100 Mbps connectivity
- 100 milliseconds latency (200ms RTT)
- Layer 3 network connectivity
VMware Stretched Cluster Benefits
- VMware Stretched Cluster allows you to achieve a disaster/downtime avoidance which is your key requirement. If it is not your requirements, then less costly solutions might be part of your design.
- Disaster recovery might happens the fastest way in this scenario because it is actually a single HA cluster which is stretched across two geographical locations (usually two offices, two datacenters).
- Disaster avoidance can be done proactively. When both locations are more far away, it might be possible to simply vMotion over distance to the remote site in order to avoid disaster and outage. In this case, you can be proactive and non-disruptive.
- A Split brain scenario where only one side of the Stretched Cluster continues to allow writes to the stretched storage. Witness component monitors whether the failed site comes back online, to allow resyncing the data between both sites.