Unfortunately, I am leaving San Francisco a day early to be back home for some meetings. So, no Alabama Shakes or trip to At&T park for me and I also know I am missing some really good sessions. That said, to wrap up my thoughts, I wanted to focus in on resiliency and availability. This is a topic we at Stratus have a lot of experience in and I will try to put it in the context of what VMware offers. As is the case with any consumer products, you will see many options out there and each will have their pro’s and con’s.
This is not meant to be comparison per se, but more of an overview of the trade-offs associated with different situations of VMware availability.
- Basic Failover – If you have a need to protect an environment that can suffer some downtime and does not require the failover into another fault domain, there is the HA option. This is actually a really great option from VMware and is probably the most used feature of Vshpere. It’s in the box, easy to set up but it does require some additional HW though (3 servers and external storage are recommended). You also have to invest some effort in figuring out where the failover lands since that can cause performance issues or even impact other workloads.
- Advanced Failover – Things start to get a lot harder when you try to get fancy. And fancy includes Clouds or Hyper-converged scenarios. For example, what if you want to failover outside the fault domain or another site? What about the underlying data store? Do I need to use VSAN? What about DRS? As you may imagine when the boundaries to placement are lower but the complexity gets a lot higher, you may need extra features from VMware and it may involve more planning.
- Fault Tolerance – Interestingly enough, in VMware the simplest option to plan and set up is FT. It also provides the most robust protection of your workloads and data. You remove the need to plan out the placement and landing of the failed over VM. You take the advanced features out of the equation and it’s easy to set up. There are some capacity and performance limitations but for the right workload it’s a great approach.
So, VMware has a lot of different options and each has its own benefits and issues. The trade-offs are cost/simplicity/performance. All need to be considered when deploying your workloads.
At Stratus we have a slightly different approach. We have hardware that fully supports Vsphere, that is also fault-tolerant, is not complicated to set up and costs about the same as 2 servers (not 3 servers and an external array). Think of it as the simplicity and robustness of VMware FT without the costs.