Listen to the entire podcast series about Red Hat’s chances with its storage server software
GlusterFS storage software will turn the market 'on its head'
With push into software-based storage, Red Hat will help drive down cost of storage features
Red Hat Inc. can approach open source storage software in the same way it did with Linux in enterprise IT shops, but the company will soon discover the differences between the two markets, according to one leading industry consultant.
Marc Staimer, president at Beaverton, Ore.-based Dragon Slayer Consulting, said Red Hat's open source software-based storage has a shot at success in the secondary storage market for passive data, but it will struggle to gain mainstream acceptance for primary storage of active data. Staimer said enterprise storage users prefer to buy their software and hardware from a single vendor to avoid the sort of finger-pointing that can happen when problems arise.
Can Red Hat bring about mainstream acceptance of software-based storage, in other words, storage software that doesn't require special hardware?
Marc Staimer: You're really talking two different paradigms. In the Linux side of it, or the operating system side of the world, you had a customer base that is used to buying software and then putting it on hardware, in x86 especially. Microsoft was and still is one of the top two operating systems -- with Linux being the second -- but you buy the hardware and you put the software on it. Actually, the hardware vendors are doing it for you; but, generally speaking, that's what you do. You accept different support for the software and the hardware.
Storage has never been that paradigm when it comes to primary active storage, and I'm going to differentiate between active and passive, passive being secondary, because passive storage is a different animal. In active storage, it's one of those things where the support in the software and the support in the hardware, people tend to want it from the same vendor, not multiple vendors. And it's because storage is a different paradigm, and they expect different things from their vendor. They expect that it's going to work, and when it doesn't work, they want one phone call to make not two. They don't want the hardware guy to go, 'It can't be us. We work fine. Go call your software guy.' The software guy goes, 'It can't be us. Go talk to your hardware [guy].' Different paradigm.
So, I don't see them necessarily in active storage bringing on mainstream acceptance, where you'll roll your own. 'Buy the software from us. Go get commodity servers and JBODs, and you'll be ready to go, and you'll have a good solid storage system.' I just don't see that happening in the mainstream. Too many things can go wrong with SAN and NAS [network-attached storage] storage when you do that.
If you're looking at it for passive storage, secondary storage, that's already taking place. People are already accepting you can roll your own because it's not your primary data. It's not the data that you're accessing every day, every hour, which you're running your business on. It's typically your backup data, your archive data, your old data. So, if you have an issue, it's not an urgent issue, and you can deal with two vendors in the mix, or more than two vendors in the mix. So, that's part of the issue of why I don't see them being able to turn this into mainstream, and other vendors in the software space have found they get much more success when they work with OEMs, people that are VARs, that package their software on hardware and take it to market as a storage system.
Do you think Red Hat can commercialize GlusterFS in the same way it did with enterprise Linux?
Staimer: Again, I'm going to separate the active data market and the passive data market, secondary data storage. In secondary data storage, that's a good opportunity for them, and yes, they can. It's a fine product for that, and it will do well in that space. But for primary, they'll need partners who actually package it and bring it to market, so it's shrink-wrapped in hardware. Shrink-wrapped in software, in a box, in plastic, is not a strong market presence. It's not something that most people will say, 'OK, ship that to me. I'm going to buy this hardware and software and then have to deal with the issues that go along with that.' I'm not saying it's impossible. I'm not saying it can't happen. But I just don't see it being very likely.
To which vendors and which types of storage will the Red Hat Storage model pose the greatest threat, assuming the software matures?
Staimer: It will pose a threat to secondary storage, so the [EMC] Isilons of the world, NetApp to a point, when they're selling their storage for secondary storage, the second tier of some storage systems. But they're trying to compete with the object storage players as well. It's going to be competitive to that space, and then it comes down to pricing and packaging at the end of the day.
Do you think Red Hat can do for OpenStack, and especially OpenStack Storage, what it did for Linux?
Staimer: OpenStack Nova [also known as OpenStack Compute], yeah, I would say it can. That's a pretty solid product. OpenStack Swift and OpenStack Cinder -- Cinder is an interface to storage systems -- when we talk about OpenStack Storage, I'm going to deal with it from a Swift perspective. Swift is object storage, and it's pretty weak object storage, and it's pretty unstable code right now. I've talked to a number of users of Swift, and I have not been impressed yet. It may be several years before it's a really strong product. This is different than Linux, too. The Linux open source community was a strong open source community before Red Hat decided to become a distribution for it and support organization for it. [With] Swift, there are a lot of people who have jumped on the bandwagon, but you don't see the stability of the code. You don't see the communities behind it at this point. I don't see where Red Hat's model is going to change that significantly. It needs to become a much more powerful, more stable, more feature/function-rich product before I would say that their model makes a lot of sense.
To achieve success with OpenStack, what will Red Hat need to do moving forward?
Staimer: Typically, for open source software to be successful, the code has to be stable. It has to solve real market problems. And it has to have the support of the general community, not just the distributor of it. An example of it would be Hadoop. Hadoop is very stable. It's got a very solid, strong management around architecting the releases of Hadoop. It's got good community support, and it's got good vendor support. OpenStack Swift has got good vendor support; I wouldn't say good community support. The code's not stable. And the management of the open source community needs to become stronger. So, if I were Red Hat, I would jump in with two feet and try to control the deliverables, the roadmap, the input from the vendors, the disciplines that have been lacking. I would be trying to impose that, and I believe they are trying to do that. I give them credit for trying to do that. That's why I say it's not hopeless on Swift, [but] just that it's going to be at least two to three years [before things happen].