OpenStack is an incredibly popular technology these days, but contrary to popular belief it is not an alternative to VMware, nor is it the final solution to avoiding vendor lock-in — even if that may have been one of the early, albeit misguided, goals. In fact, VMware is one of the top contributors to this open source project and the real appeal of OpenStack is the API that allows developers to build the type of modern, scale-out apps that have become popular on Amazon Web Services.
Since OpenStack is fundamentally an API for consuming cloud computing resources, the real question facing datacenter architects today is: what resources should be made available for consumption? Thanks to efforts from VMware, along with other contributors, support for vSphere has substantially matured and is now a strong alternative to KVM, which was popular initially due to the open source nature of this project.
According to a recently-published performance study, an OpenStack infrastructure based on VMware technology is faster and less expensive than an equivalent stack built from Red Hat products, concluding:
- VMware VSAN delivers 159% more IOPS than Red Hat Storage Server (GlusterFS)
- A 16-node Cassandra NoSQL database performs 53% better on vSphere than on Red Hat KVM
- The total cost of infrastructure hardware and software is 26% lower on VMware than on Red Hat
Thanks to the vSphere/VSAN hyper-converged infrastructure, there is no need to build dedicated clusters of shared storage like there is with GlusterFS. Instead, shared storage functionality is provided by pooling disks and SSDs that are directly attached to hypervisor hosts. This offers capacity and performance for a range of applications as well as flexible redundancy options — administrators can configure policies to accommodate one or more replicas of critical data across the cluster or even opt for no replication on non-essential workloads.
In addition to better performance at a lower cost, VMware also delivers a platform that is suitable for all workloads — not just design-for-fail cloud applications. Applications running on a vSphere cluster benefit from VMware HA and DRS, reducing downtime and increasing performance as VMs are balanced according to shifting resource demands. Don’t forget that the purpose-built ESXi hypervisor also has much broader guest operating system support and a smaller attack surface that translates into reduced host downtime for patching and maintenance.
Now you can give your developers the agile cloud API offered by OpenStack with the trusted reliability and proven performance of vSphere — the best virtual infrastructure for any application.
Not really an objective outcome of this performance test don’t thing. I always have a funny feeling about vendor commissioned performance tests.
I this case it underlines this gut feeling with the fact that they have tested Glusterfs which is never positioned as an optimal Openstack Storage platform. That’s why Redhat acquired Inktank for it supported Ceph releases, I guess VMware’s claims don’t stand against a similar product.
Speaking of vendor-commissioned performance tests… Red Hat already went through the trouble of proving that GlusterFS was over twice as fast as Ceph for OpenStack clouds. It is only logical to test against GlusterFS, not Ceph.
Read all about it here.
Thanks for your comment. In the future, please disclose your vendor affiliation.