Storage Q&A: How to run Virtual Machines on your OpenStack Infrastructure

Open vStorage has a new product that will allow you to run virtual machines right on your OpenStack Infrastructure. Storage Crump Founder and Lead Analyst George Crump and Open vStorage Product Manager Wim Provoost sat down with me in the studio for a podcast to talk about Open vStorage.

Charlie: George let’s start with you, before we talk about why and how to run virtual machines on OpenStack, can you explain exactly what OpenStack is.

George: We could probably do a whole podcast just on OpenStack. Essentially it’s a set of software tools for bundling and managing computing platforms for public and private clouds. It’s an open standard, which makes it much easier to adopt than other closed infrastructures, and it’s backed by some of the biggest companies in software development as well as hosting companies such as Rackspace and IBM and HP. Really big initiative in the market today.

Charlie: Wim, why you would want to run VM on an OpenStack infrastructure?

Wim: There are many reasons, so lets start with the first one. Just as an example, OpenStack is the only solution that allows mixed hypervisor environments, which means you can mix different kinds of hypervisors within one environment. This is becoming more and more important as IT is becoming more fragmented over time.

Open vStorage offers almost the same technical capabilities. For example: migrating virtual machines without down time just like any proprietary alternative, but at a fraction of the cost because it’s Open Source. As there are many vendors in the OpenStack ecosystem, the chances of getting locked in by a vendor is also slim to none.

Charlie: And I assume that Open vStorage is the missing ingredient that can make this happen. Sso what is Open vStorage?

Wim: Open vStorage is a storage platform. This means that Open vStorage is a middle storage layer in between virtual machines and one or more storage back ends. It kind of extracts the backend from the virtual machine and creates a unified view on that storage.

It also uses the flash inside the host to act as a cache, we have read cache, we have write cache. When data is no longer needed inside the cache, it is pushed to the storage backend where it is kept for long time retrieval. This backend can be almost anything: SAN, NAS, object storage, or a distributed file system. Lately we even added support for Ethernet drives just like the Seagate kinetic drives.

Charlie: What problems are Open vStorage trying to solve?

Wim: Open vStorage is trying to solve a really big problem, making object storage more useful. Up until now you haven’t been able to use it as a primary storage platform. So running virtual machines directly on top of your object storage is basically impossible. There are many issues that are preventing that use case. For example: eventual consistency and object storage. While they are built on large disks so they have a high latency and low performance, virtual machines actually require low latency and high performance storage, so it’s quite the opposite. There are also management paradigms. For example hypervisors require block storage while object storage requires objects, so they are like cats and dogs and they really don’t get along with each other.

Charlie: George, I know we’ve written a lot about object storage on Do you agree that there’s a role for object storage to support workloads that traditionally expect block based storage?

George: I do. It is true that most object storage systems are designed to store a lot of objects i.e. files, so we think of it for unstructured data. But a lot of the basic capabilities of object storage, like it’s ability to be very durable with data, to be able to survive multiple drive and node failures, to be able to scale. All those different things are just as important to block based storage systems as they are to unstructured data types.

But really, as Wim has mentioned, the missing ingredient is how to make that bridge, how to make object storage act like block storage, and these are the capabilities of products like Open vStorage.

Charlie: Why does Open vStorage focus on the OpenStack market?

Wim: First of all, OpenStack is really booming. We have the conference coming up in Paris, and you can already see that everyone is going there so it’s really taking off. But we at Open vStorage see that theres a problem that we can tackle. For example, if you want to use the object storage which is native in OpenStack, (called Swift) you can use it to store your virtual machine images and backups, but you can’t use it to run virtual machines.

What we’ve developed is an OpenStack cinder plugin to easily integrate OpenStack Swift as a primary virtual machine storage platform. Which means you can directly run virtual machines on top of this object store. This also means that you don’t have to maintain two separate storage platforms, for example one for block and one for object storage.

Charlie: Is Open vStorage trying to replace CEPH as OpenStack storage?

Wim: No, not at all. We’re actually complementary to CEPH, we support CEPH as a storage backend like we support Swift or other storage backends and other object stores. But by using Open vStorage in combination with CEPH we believe that we can enhance CEPH and make sure that it becomes an even better product than it already is.

To give you an example. CEPH has an erasure coding built in, but the CEPH block interface produces really small blocks so erasure coding makes no sense on small blocks, and the only option you have is to replicate the small blocks, which of course you’re using roughly two times the amount of storage hardware to store the actual data. Next, Open vStorage uses an advanced caching system inside the host, the KVM host, and we believe that this will lead to better storage performance, compared to running natively in CEPH. Of course, Open vStorage is capable of being the middle man for iSCSI and CEPH, because currently iSCSI and CEPH cannot be integrated directly.

Charlie: Why did you guys decide to go Open Source?

Wim: We firmly believe that storage is too important to keep proprietary. Open Source is used to follow commercial closed source offerings, but lately it’s the other way around. Open Source is cutting edge. Its leading the way. The reason for this is quite simple. More eyes with different looks are looking at the codes, and we believe that if you have more people looking at the codes it will become more innovative.

Of course you have a community that helps you and turns it into a product that is needed in the market. And we believe that if you turn it into a paid product people will pay for it because they like Open Source, and they don’t like the hassle of setting everything up and doing the monitoring. So that’s why we went Open Source and we still believe that there’s business behind it.

Charlie: George, does the Open Source model make sense here?

George: Its interesting, Open Source has always had a dichotomy of a challenge for several reasons: you want to take advantage of the population, or the community, that can help innovate and keep it going, you want the lack of vendor lock in, and you want to keep costs down. There’s a lot of reasons why Open Source type of products make sense. The challenge is converting that into a business model that you can afford to do things with. I think that we clearly have a track record now of companies that are doing that.

As we move into enterprises and maybe mid tier and large cloud providers, these guys are not going to want to spend the time working on all these things, they’re just going to want to be able to install something and have it work. A lot of times, if your business is to not necessarily deliver an IT service but to deliver something else, you don’t want to spend a lot of time on the IT minusha. And so leveraging capabilities like Open vStorage makes a lot of sense, but at the same time you get the benefit from the backend and all of the Open Source community. So yes I think there is a business model here and I think it makes a lot of sense, and now it’s just a matter of execution.

Charlie: So what is the business model behind Open vStorage?

Wim: Well, currently theres a mantra in the IT industry: if you have money you go with VMware, if you have time you go with OpenStack. What we will do is release a converged infrastructure offering, based on OpenStack and Open vStorage. So cloud vendors, the sponsors of Open vStorage, will provide the whole stack – the hardware, the storage backend, monitoring, support – and we will even set up OpenStack. We will do this for channel partners who don’t have their own resources or knowledge to set up and maintain their own Open Stack infrastructure or a private cloud.

Charlie: So George, what is the StorageSwiss take on running VMs on a Open Stack infrastructure in general, and Open vStorage in particular?

George: I think that Wim nailed it with the statement that if you have money you go with VMware, and if you have time you go with OpenStack. That’s a great way to sum up the market. But what if you don’t have a lot of time and you don’t have a lot of money? Then you have to look at something that will help you cross that bridge, and I think that products like OpenStack with Open vStorage are a very viable way for people to get there.

The challenge in the past has been getting these more turnkey products, because people don’t have time to tinker around in most cases. Getting these bridge products is critical. And I think what we’re seeing here with Open vStorage is one of the first products to allow you to run business class type of applications, backend type of applications, in the Open Stack infrastructure. So I think there’s a lot of value here. As Wim has also said, Open Stack is booming and getting a lot of interest and that makes it more approachable to the non-tinker type of person.

Listen to the podcast now

Click Here To Sign Up For Our Newsletter

Tagged with: , , , , , , , , , ,
Posted in Storage Q&A
2 comments on “Storage Q&A: How to run Virtual Machines on your OpenStack Infrastructure
  1. Tim Wessels says:

    Well, I don’t think I noticed it in the interview, but Wim Provoost and his company CloudFounders would be smart to take Open vStorage and put it on a hardware appliance running OpenStack and sell it under their name. And if you take it one more step, go buy some Cloudian HyperStore hardware appliances for your S3 compatible object storage, which is capable of serving the purposes of the Open vStorage router. Rack them all together and you could have a “turnkey” OpenStack solution that might take a couple of days but not weeks to get up and running. The hardest part is clearly the orchestration of the OpenStack programs. I recomend reading my colleague, Paul Venezia’s, recent review for InfoWorld of the Mirantis OpenStack distribution that uses Fuel to do the management and orchestration heavy lifting for OpenStack. Sorry I missed Wim when he spoke at the OpenStack Meetup group at Micrososft NERD in Cambridge, MA, on January 14th. It was just too darn cold to make the trip into Cambridge from the woods of southwestern NH where I live. Wim, please come back in the spring or summer or fall when New England is much more hospitable.

  2. Hi Tim,

    thank you for the kind words. You missed an excellent Meetup in Boston and yes, it was cold, very cold outside. The Meetup in Washington DC had the same content, was taped and placed online .

    Yes, we will create an converged Open vStorage solution based upon OpenStack as you hinted. We are not really fans of hyperconverged as discussed in my latest blog post ( The goal is to create a turnkey cloud solution based upon OpenStack. With Fuel you can indeed set things up very easy. The biggest headache for OpenStack is not networking or setting things up, it is now storage. Without Open vStorage you need to go for expensive SAN storage or you need to be happy with underperforming storage. That is what we wanted to change with Open vStorage.

Comments are closed.

Enter your email address to follow this blog and receive notifications of new posts by email.

Join 22,221 other followers

Blog Stats
%d bloggers like this: