RHEV 3.1 storage: Functionality and considerations
Virtualization expert Sander van Vugt explains how RHEV storage works and discusses some of the new storage features in version 3.1.
Red Hat Enterprise Virtualization (RHEV) is an open source virtualization platform that can perform the same functions as hypervisor competitors such as VMware or Microsoft Hyper-V. With the latest version, RHEV 3.1, Red Hat added live migration that allows admins to move a running virtual machine from one SAN to another without downtime. Other storage features in RHEV 3.1 such as snapshots have been added that are making it even more relevant in the market .
In this podcast, Linux high availability and virtualization expert Sander van Vugt explains how RHEV storage works and why he believes Red Hat is gaining more footing in the virtualization world. Listen to the podcast or read the transcript below.
Can you start off by giving us an overview of what Red Hat Enterprise Virtualization is?
Sander van Vugt: Sure. Red Hat Enterprise Virtualization is providing a management interface, basically for KVM [Kernel-based virtual machine] virtualization. And KVM is Linux virtualization, and you can compare that pretty well to what VMware is doing with ESX. KVM by itself is like a free version of VMware ESX. It's just a standalone hypervisor, which does a great job, but in order to do the job a little bit better, especially at an enterprise level, you need a decent management interface. So from the open source community, there wasn't really anything decent yet that could be used to manage several KVM virtualization hosts.
So Red Hat has first sponsored the Ovirt product -- that's an open source product and is used as the input product for RHEV. The result is a Web-based management interface that is based on the Tomcat server, and that can be used to integrate multiple KVM hosts and perform tasks like live migration and high availability in a smooth and easy, and very affordable way. So that's basically what Red Hat is all about.
How does storage function differently with Red Hat than it does with VMware or Hyper-V?
Van Vugt: One thing that is quite unique with RHEV is that Red Hat storage is added. Now let me explain exactly what Red Hat storage is doing. In normal virtualization solutions, storage is mostly on the SAN, which means that there is a centralized device, and on the centralized device, you will create a disk, and you will share that between different hypervisors. So every hypervisor basically is writing to the same disks on the same SAN. It doesn't really matter if your SAN is redundant because even if it is redundant, it's still the same disk.
Now with Red Hat storage added, storage can be allocated on different machines, and Red Hat storage decides exactly where the data is stored. This is a very clever way of creating a distributed file system, which makes sure that virtual machines are stored in the data center where they really are needed. So to summarize, the difference of how storage is handled in Red Hat as compared to VMware, for example, is the decentralized storage approach.
What are some storage-related functions that are in 3.1, and what do you think is the most important?
Van Vugt: In Red Hat 3.1, the most important storage feature is definitely live migration. The idea [behind] live migration is that there's this data center, and the data center in previous versions of Red Hat was the basic unit that was used to create a Red Hat environment. That means that hypervisors are associated to a particular data center, and a data center is connected to one particular kind of storage. In previous versions of RHEV, it wasn't possible to move storage from one data center to [another] without bringing down the data center. And that means if you have a hypervisor using one particular data center -- and the storage on that one particular data center -- in a migration scenario you need to bring the hypervisor over to another data center. That's something that's really hard to do, and uptime is really important to you. Because the only way to do that in 3.0, for instance, was to bring down your entire data center and move over the storage to the other data center. Now, fortunately in version 3.1, you can do that without any downtime, and I think a major problem that existed in previous versions of Red Hat has now been solved, which makes the product ready for the big data center environments.
How do you think RHEV 3.1 does with scalability of storage?
Van Vugt: With regard to storage, scalability is mainly improved by adding Red Hat storage. Because Red Hat storage, as I explained, is the solution to make sure that storage is available at the most broad scale and not really tied down to one data center or one physical location. Red Hat storage is the solution that is really virtualizing storage, and that makes it more scalable.
Red Hat came out with its Red Hat Storage Server storage product recently. Would you consider that to be software-defined storage, and how would you define "software-defined storage?"
Van Vugt: Software-defined storage -- that is something that was invented by marketing people. It's one of those terms that I consider a little bit difficult to grasp. Now, you could consider it software-defined storage if you would say that software-defined storage is the storage interface that doesn't occur directly to one particular hardware device. If that is what you mean by "software-defined storage," yeah, Red Hat storage definitely is software-defined storage. But, in the end, it all sits on real hard drives in real SAN environments on real servers.
How would you say Red Hat's growing integration with the cloud affects storage for that company's product?
Van Vugt: Red Hat is focusing on cloud pretty heavily now, they've got a cloud-form solution that they are using that attempts to get into the provider world, and [with] those kinds of environments, you need flexible storage. From that perspective, the way they are handling storage and extracting storage, and [the way] they are virtualizing storage in 3.1, really makes sense in the cloud environment.
This is something that cloud providers are going to understand, and they are going to like it because if for some reason they need to move over their infrastructure between data centers, Red Hat storage is really going to help that.
Do you think there are any storage functions that are missing from RHEV 3.1?
Van Vugt: Well, honestly, I'm sure that in some upcoming version they will introduce something that I wasn't aware of that I needed right now, but for the moment, I'm pretty happy with the storage functionality they are providing. And there isn't really anything obvious that I would say, "Hey, this is missing," to make RHEV a solution that can compete with Hyper-V or VMware, for instance.
Do you see RHEV becoming more competitive with the other virtualization products in the future?
Van Vugt: Yes, absolutely. In my opinion, there are three major players on the market right now. [There] is VMware, because they really make a very good virtualization solution, but it is expensive. There is Hyper-V. Hyper-V is gaining ground because it is Microsoft, and if a company is using a large Microsoft-based infrastructure, it just makes sense to use Hyper-V. And now there is Red Hat Enterprise Virtualization, and RHEV pretty much comes to the level where all the basic functionality that is provided by VMware is provided in Red Hat as well, but it is a lot cheaper than VMware.
Between Hyper-V and RHEV for companies that didn't make their choice yet, I think RHEV can match Hyper-V for sure. And for VMware, for those companies that already have large VMware-based infrastructure, of course they are not doing to move away from VMware overnight and replace their huge investment base in VMware by a Red-Hat-based infrastructure.
For new parts of the infrastructure that need to be virtualized, and especially if those new parts are Linux-based servers, RHEV makes sense. And I already meet customers that tell me, apart from their VMware infrastructure, they're just going to mount Red Hat infrastructure because their Linux admins like working with solutions that are Linux-based, and that is so much closer to what they are doing in their everyday work.
So to give you a short answer, RHEV is not going to wipe out VMware, but it is getting a solid place in the enterprise environment beside VMware. If you go to the new markets, markets that haven't virtualized yet, if those customers aren't afraid of Linux, Red Hat is much more affordable than VMware.
What do you think is the most important thing to keep in mind when choosing storage for RHEV?
Van Vugt: The most important thing that people should keep in mind is that in the last release of RHEV there hasn't been much attention to the integration with Red Hat storage. We have talked about Red Hat storage a lot in this interview, but it's not like it's the only storage solution, and not all storage should be Red Hat storage-based because in RHEV you can still integrate your normal SAN environment. Everyone should keep in mind that for most environments, especially if it is a single-site, the basic SAN-based storage solution is still the way to go. It's not as fancy as Red Hat storage, and not as advanced, but it is a decent, solid way to organize the way that storage is handled in those environments.
Sander van Vugt is an independent consultant specializing in Linux. Based in the Netherlands, he has authored a number of books related to Linux including Beginning the Linux Command Line, Beginning Ubuntu Server Administration and Pro Ubuntu Server Administration. He is also a frequent contributor to SearchVirtualStorage.com and speaker at Linux-related conferences around the world.