Many IT professionals consider their VMware infrastructure to be a private cloud. While not true self-provisioning, it provisions compute and storage resources much faster than in the past. Public cloud providers like Amazon and Azure have almost unlimited quantities of these resources so the thought of leveraging them is appealing. IT professionals want to create a hybrid cloud that utilizes their on-premises cloud and the public cloud. The decision for IT professionals to make is how exactly to move VMware to the cloud. There are several options, each with their own pros and cons.
Why Move VMware To The Cloud?
Most organizations are not looking to move their entire VMware environment to the cloud and have it stay there. Instead, they are looking to leverage the cloud as a disaster recovery site, to burst specific workloads into the cloud, to leverage cloud storage capacity or to run Test-Dev simulations. But the intent is for the primary data center to remain the primary data center.
How to Move VMware to The Cloud?
As appealing as the cloud may be to IT professionals, they need to fully understand how they will move workloads to and from the cloud. While both Amazon and Azure announced native support for VMware, the challenge is getting data to the cloud and keeping that data updated. Most approaches require replicating data from the primary data center to the cloud provider. After the initial data copy is complete the replication process will continue updating the cloud copy to make sure the cloud has as current a copy as possible.
The replication approach creates a couple of challenges. First, data is now in two locations and IT is essentially managing two data centers. Second, there is the transformation from a replicated copy to a instantiated image. While the instantiation is made easier by native VMware support, there is still the process of stopping replication, making a usable copy of the virtual machine (VM), starting that VM, and then restarting replication.
IT professionals also need to be careful as to which class of storage they are replicating to. As we discuss in our recent column “Understanding The AWS Storage Portfolio” Amazon, for example, has several different options. They have cost effective targets that aren’t suitable for production data and they have high performance targets, that are suitable for production, but expensive.
If the cloud is being counted on for disaster recovery or cloud bursting, that cloud VM may make unique changes to production data. Those changes need to be very carefully synchronized back into the on-premises data set.
Another option is to centralize the data set in the cloud, essentially using the cloud as primary storage. This approach simplifies data management and reduces on premises storage costs. But it must overcome the cloud latency problem. An on-premises appliance that caches data can help, but for applications the implications of a cache miss, even if it is rare, is too great. While a file system might tolerate seconds of delayed access, an application will likely crash.
To solve this problem, the cloud as primary storage approach needs to combine an on-premises appliance with a regional data center acting as a second cache before data gets to the cloud. The regional center can respond to an on-premises cache miss quickly enough as to not cause a application crash but still make sure that all data is automatically and safely stored in the cloud.
It’s great that the public cloud providers are delivering native support for VMware. The question still remains though what is the best way to get data to the cloud. Which approach is best will vary by organization. To determine which is best for your enterprise, watch our on demand webinar, “How To Leverage Cloud Storage for Hybrid VMware”.