Red Hat Storage Server seeks to mimic success of enterprise Linux
A comprehensive collection of articles, videos and more, hand-picked by our editors
Listen to the entire podcast series about Red Hat’s chances with its storage server software
Red Hat's open source software-based storage holds appeal mainly for passive data
With push into software-based storage, Red Hat will help drive down cost of storage features
Red Hat Inc. will eventually turn the market "on its head" with its Red Hat GlusterFS-based storage software in the same way that it successfully challenged major server operating system vendors with its distribution of enterprise Linux, an IDC analyst predicted.
Ashish Nadkarni, a research director in the storage systems practice at Framingham, Mass.-based International Data Corp. (IDC), said Red Hat may need a few years to realize its vision of converting commodity hardware into a full-fledged storage platform.
But Nadkarni thinks Red Hat Gluster will challenge scale-out, file-based storage systems from major vendors as well as combined compute-storage platforms that cater to the high-performance computing market with data analytics software.
In this podcast interview with TechTarget's Carol Sliwa, Nadkarni also offered predictions on Red Hat's chances to help drive acceptance of the open source OpenStack compute and storage platform, and discussed the changes that will need to happen for Red Hat to achieve success in the storage market.
Can Red Hat bring about mainstream acceptance of software-based storage, in other words, storage software that doesn't require any special hardware?
Ashish Nadkarni: I believe it can. If you look at what it has done with Linux, it has successfully transformed Linux into an operating platform that can today run on any x86-based hardware from any vendor. It can run as a virtualized platform. It can run as a cloud instance. And if there is such a thing as a software-based operating system, then that is Linux. And in the enterprise, Red Hat Linux is the most commonly used version of Linux, and if you look at the deployment diversity, it speaks for itself. And if you take software-based storage, the premise, the goal and the mission for software-based storage is to do exactly that: Take a general-purpose operating system, convert it into a purpose-built storage platform, and leverage the resources that you have available to you internally inside of a server and use those for creating a storage platform that can offer all of the storage services, so you can take any commodity hardware and convert it into a full-fledged storage platform.
Do you think Red Hat can commercialize GlusterFS in the same way it did with enterprise Linux?
Nadkarni: In principle, I believe it can commercialize GlusterFS because if you look at what happened when Red Hat embarked on a Linux journey, the big sharks in the ocean were the incumbent operating platform system vendors such as Sun Microsystems, IBM and Hewlett-Packard. Even Microsoft was in the game with Windows Server. And Red Hat was the underdog, and it started in with a very small share of the market and essentially grew it into a bigger share. They did not really go after a new market. They took an existing market and turned it on its head, and they are doing the same thing today with GlusterFS. I would give them a few years before they are able to successfully turn the thing upside down.
Along the way, they are going to get a lot of headwinds from the incumbent storage suppliers like EMC and IBM and HP -- pretty much the same cast of characters that were in the game before. EMC is an exception, but everybody else was also in the game with their operating platforms and had pretty much done the same thing at that time. And you will see that might play back again -- history repeats itself. However, the one good thing in this situation is that Red Hat is not trying to go after mission-critical or business-critical environments for storage systems today. It is trying to go after the low-hanging fruit, which is the unstructured data, semi-structured data, data sets that require cheap and low-cost storage systems and platforms. Today, if you go to try and store unstructured data on a commercial storage platform, companies will end up paying a lot more. So, I think they're going after it the right way, but they are obviously going to get a lot of opposition and headwinds from the incumbents.
To which vendors and which types of storage will the Red Hat Storage model pose the greatest threat, assuming the software matures?
Nadkarni: Before we go there, let me briefly talk about what Red Hat's platform is. Red Hat's platform in IDC's world is software-based, scale-out, file-based storage. I know it's quite a mouthful, but what it comes down to is that it can run entirely as a software solution either on a hardware platform, meaning it runs bare metal; it can run inside of a VMware environment; or it can run as a cloud instance. But, fundamentally, it is a scale-out, file-based storage platform. What that means is that the first group of vendors that it is potentially going after is the scale-out, file-based storage vendors. If you look at that market, it is the EMC Isilon, the HP Ibrix, the Dell FluidFS or the IBM GPFS -- all of these scale-out platforms that are based on distributed file systems. Those are the platforms that Red Hat is seeking to compete with.
The second group of vendors are these compustorage vendors, as we call them, where the compute and storage platform itself are being fused together in order to make it efficient for the workload to access the data. If you look at the startups in this space, there's Nutanix and the Pivot3s and the Scale Computing and SimpliVitys of the world. They are definitely in Red Hat's sights. Why? Because Red Hat has Red Hat Enterprise Virtualization, a hypervisor platform built on KVM, which it is trying to integrate with GlusterFS so that it can create a compustorage or a natively converged infrastructure, which is low cost of course.
And the third group of vendors Red Hat is going after is these [high-performance computing] HPC/analytics vendors. With a Hadoop connector for GlusterFS, it is seeking to allow folks who have lots of data to run Hadoop workloads natively on GlusterFS. And similarly within the custom HPC world, there's a lot of these distributed file systems that are used for HPC workloads. Certainly, it would be comparable to if you run it with GlusterFS.
Do you think Red Hat can do for OpenStack, and especially OpenStack Storage, what it did for Linux?
Nadkarni: Yes. What OpenStack lacks today is companies like Red Hat and, to be fair, it's not just Red Hat. There is SuSE, which also took a version of Linux and converted it into enterprise Linux. Technically what Red Hat can do with OpenStack is bring about the discipline and the rigor, the standardization that OpenStack lacks today. Today, what is happening is OpenStack as an ecosystem is becoming very fragmented in nature. You have the Rackspaces of the world -- service providers who are downloading OpenStack and customizing it for their own delivery purposes. Then there are the HPs and the IBMs and the Dells of the world who are customizing OpenStack in order to make it suitable to be delivered using their own hardware. And then there are, of course, the software-only folks who are looking at OpenStack as a way to augment their existing solutions. If you look at all these groups of people, they are all taking an existing OpenStack distribution and customizing it for their needs. And that is creating fragmentation.
What the community needs today is a standard OpenStack distribution. If you take the OpenStack storage projects, such as Swift and Cinder and to some extent Nova, because Nova also is a compute integration with KVM, and you look at this one big spaghetti mess that is happening, it needs companies like Red Hat. If a business wants to deploy OpenStack, who do they go to? Most of these businesses who are Linux shops need not look any further. They can go to Red Hat and say, 'Hey, we are familiar with Linux. We know what you have done for Linux. We are familiar with your support model. So here you go. Why don't we give Red Hat's OpenStack distribution a shot instead of trying to spend hours and hours trying to download the free version and try and customize it?' And in the process, if something breaks, then the business is on their own, whereas with Red Hat, they can do essentially what they did with Linux: Buy maintenance from Red Hat and essentially use the same support structure that they're used to.
To achieve success with OpenStack, what will Red Hat need to do moving forward?
Nadkarni: Red Hat is one of the most prolific contributors to the OpenStack project, and that's big. When you look at this ecosystem where everybody claims to be an OpenStack provider or supplier or contributor, ultimately what it comes down to is who is contributing the most amount of code to OpenStack and who is really investing their R&D dollars in OpenStack. The amount of contributions that Red Hat is making needs to be made public. People need to realize that Red Hat is really investing its own dollars to create that standardization in terms of distribution and such. I think Red Hat needs to do more to get this information to businesses, especially businesses that are evaluating OpenStack and that may already be Red Hat customers.
The second thing it needs to do is to prove that it is not using OpenStack as a convenience thing to push its own storage platform and, in fact, it is really committed to OpenStack. The more standardization that comes around with OpenStack, the better it is for the community, not just for Red Hat. It's about trying to show to the businesses out there today, to the market that Red Hat's commitment to OpenStack is similar to what its commitment is to Linux, and ultimately that translates to a standardized platform for businesses that are on the fence because of some of the challenges that they've heard about when people try to deploy OpenStack.