If the Broadcom acquisition of VMware is causing you to start to rethink your application deployment strategy, then you have been missing a key cost driver in your infrastructure.
IBM tests show that transaction workloads on x86 can be delivered at over four times the throughput and in half the response time using a Red Hat® OpenShift® environment versus a classically deployed virtual machine environment.
In other words, take the hypervisor out of the resource path, and you get to run 4x more workloads on the same hardware capital investment.
A little more detail from the IBM paper:
It takes (4) x86 servers running VMs to do the work of one like server running OpenShift. I encourage you to read the source paper referenced above – only 6 pages of content – may want to read it twice to assimilate all the facts – a lot of details.
Now the reality is - to recognize all these benefits requires the application to be refactored to run natively in containers. The application used in this study is WebSphere which comes in both Virtualized and Containerized versions. IBM is working to refactor many applications to take advantage of containerization and pass those benefits to our customers. Another example is Fusion Backup 2.6, which was originally designed to protect Virtualized workloads. It has been rewritten to run OpenShift native and protect containerized workloads. As noted in the Fusion reference, the conversion saves customers about 2/3rd of the CPU resources and about half of the memory to protect their container workloads with improved functionality from the virtualized version.
Refactoring applications that were originally engineered to run on physical or virtual machines is a software engineering task – and depending on how well the application was written to begin with, can be a sizable task. Both IBM and Red Hat offer help for determining the fit for an application to be refactored to micro-services. But this is where OpenShift Virtualization comes to help. Some applications may be easy to refactor, some (like WebSphere) are already available in Container native form. Windows and Linux VMs can be run in OpenShift containers. This is not as good as refactoring. You won’t get all the resource savings, but it does help you on your journey to cloud native infrastructure. If you are looking for some help writing or re-writing your applications, you should check out watsonx Code Assistant.
Oh, by the way – OpenShift containers come with Linux licenses, so you may save the cost of the OS license in the VMs.
How do you eat an elephant? One byte at a time. Your legacy virtualized application infrastructure is the elephant in the data center room. The acquisition of VMware by Broadcom may turn your elephant into an elephant sized albatross burdening your budget plus your capital assets and your operations.
It will take time (years) to transition from VMs to containers. It will be a gradual transition running hybrid infrastructure with some components running in VMs and some in containers. I have talked to customers who put in a 5 year plan to make the transition, converting 20% of their applications per year. Turns out the first 20% are pretty easy, you can find the low hanging fruit and make the conversion. The next 20% are harder, through the analysis they realize there are going to be a significant number of virtualized applications that will be very difficult or impossible to refactor to microservices. So, the reality is they will be running hybrid VM/container environment for a long time.
I think it is safe to say all of these customers’ hybrid environment is Kubernetes running on Virtual Machines – a very significant part are running OpenShift on VMware. It made sense. They had infrastructure set up to run VMware (servers, storage, and networking) and Red Hat fully supports running OpenShift on VMware (including IBM Fusion). The resource inefficiencies are not as significant when dealing with sunk costs and hardware depreciation. But when it comes to sourcing new hardware and renewing software contracts, the OpenShift on VMware deployment option becomes a lot more burdensome. As mentioned above, there is an option. You can run your VMs in containers on OpenShift. Now this doesn’t mean that you can cancel your VMware/Broadcom contract. Not all VMs will run well in containers today, but many (most) can. Red Hat is driving the refinement of OpenShift Virtualization (upstream Kubevirt) hard, so that more and more VMs can run in containers. The Red Hat Migration Toolkit for Virtualization (MTV) provides tooling to help you migrate your VMs to OpenShift
I was speaking to a customer the other day that recently did a 3yr renewal of their VMware contract (ahead of Broadcom closing the deal). The conversation quickly turned to creating a 3-year plan to get all their VMs onto OpenShift via either OpenShift Virtualization or container native refactoring. (Just because you are focusing on getting off VMware, doesn’t mean you should take your eye off the value of refactoring applications discussed at the top of this blog post.)
If you were on a multi-year Digital Transformation journey to a cloud native infrastructure, you shouldn’t lose any sleep over the Broadcom acquisition. You were on the right path, though you may want to think about flipping the infrastructure stack to run VMs in containers as your hybrid infrastructure.
If you haven’t started your digital transformation journey, you can rest easy knowing that Red Hat and IBM are here to help with services and technology.
For more information on running VMs on OpenShift Virtualization, take a look at Rob Coventry’s video on the YouTube Shift Happens channel.
For additional information:
IBM Storage Fusion
OpenShift Virtualization
Animal images from Pixabay