OpenStack's prospects: Red Hat, VMware agree to disagree

OpenStack's prospects: Red Hat, VMware agree to disagree

Summary: OpenStack will either change the cloud computing universe or be a muddled mess. Red Hat and VMware execs square off.


OpenStack, a cloud platform started by Rackspace and NASA, is either the biggest virally growing open software movement since Linux or a murky hodge podge of technologies that could flop.

It all depends on what tech vendors you ask and the horses they have in the cloud architecture race.

Most of the horses---at least the vocal ones---to date appear to be on the pro-OpenStack side of the equation. Rackspace's recent first quarter results highlight the power of OpenStack. Rackspace's first quarter and outlook illustrate the power of OpenStack. Rackspace's quarter was a disappointment, but Wall Street analysts kept gushing about the power of OpenStack.

Here's a look at the Red Hat vs. VMware sides of the OpenStack equation and the reality check provided by Virtual Instruments CEO John Thompson, a 40 year tech industry vet and former chief of Symantec. Executives were talking cloud infrastructure at a Jefferies technology conference in New York.

The bullish take

Brian Stevens, CTO of Red Hat, said "only Linux has developed as virally with the community as OpenStack." Two years ago, Rackspace and NASA said they were going to build cloud infrastructure openly and OpenStack has "lit up in terms of mindshare."

"Customers around the globe have heard of this thing," said Stevens.

Note that Red Hat is OpenStack's third largest contributor. Rackspace has moved OpenStack into an independent foundation and enterprises are hot to figure out how to deliver Amazon Web Services-like clouds internally.

Stevens' take isn't hard to replicate. HP has validated OpenStack with its public cloud infrastructure. At Temple University's 12th annual IT awards in Philadelphia, Adrian Gardner, CIO at the NASA Goddard Space Flight Center, told me his organization is rolling out OpenStack at scale. NASA released contributed code, let the open developer community run with it and now is pulling in the software to scale cloud operations.

Meanwhile, dozens of technology vendors are on the OpenStack bandwagon. Stevens argued that OpenStack could become a standard.

The pessimistic view

It's obvious Stevens was rooting for OpenStack. He has a horse in the cloud platform race and Red Hat could find a revenue stream from it. OpenStack often has Red Hat's enterprise software under it.

VMware's Raghu Raghuram, executive vice president of cloud infrastructure and management, isn't so sold on OpenStack. Of course, Raghuram also has a horse in the cloud platform race. VMware sees its VSphere as the cloud OS of choice. Today, VMware is a de facto cloud standard for hybrid and private cloud shops.

"What is OpenStack? It's hard to say what the stack is," said Raghuram. "Every vendor is saying something about OpenStack and its pieces."

Raghuram's point: The last time dozens of hardware and software vendors flocked to the latest greatest play was Xen, an open source hypervisor that was trumped by KVM.

Raghuram said there are two outcomes for OpenStack. It will either develop like Linux or it'll go the route of Xen. "OpenStack needs to define clear boundaries about what it is," said Raghuram. "It's a long ways from that and a significant work in progress." He added that some customers are using OpenStack on top of VSphere.

Meanwhile, Raghuram gloated that Rackspace, the champion of OpenStack, "is using a lot of VMware" for its hosting business.

The reality check

Thompson provided a nice reality check between the Red Hat and VMware camps on OpenStack. "We don't have a dog in this hunt," said Thompson. "But profits follow architectural control. There's no way any vendor would allow a discussion about standards to go without a say."

That last comment highlights why every tech vendor and their moms are involved with OpenStack somehow.

In the end, Thompson argued that customers will push for interoperability between public and private clouds.

Ultimately, the OpenStack debate revolves around architecture control. "This industry has evolved around big profit pools controlled by a particular architecture," said Thompson. "That isn't going to change. The tech industry runs the same play every 10 years."

Stevens and Raghuram smiled when Thompson outlined his take. "Because you don't have a dog in this fight you can be more open," quipped Raghuram.


Topics: VMware, Cloud, Hardware, Linux, Open Source, Virtualization

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • The very real danger of building your sand castles on the open source beach

    This underscores the very real dangers of building critical infrastructure on systems that may be abandoned or wither on the vine due to lack of support. Sourceforge is littered with abandonware. It is equally as important to pick a solution with long-term support and commitment as it is to invest in solutions that meet your specific business requirements.
    Your Non Advocate
    • The danger exists everywhere...

      Red Hat (as opposed to, say, Ubuntu) is about as business-savvy as they come in the FOSS world. I would trust a Red Hat Open Source solution, so long as it were the best choice for whatever I need it to do.

      There are plenty of projects that wither on the vine in both the Open Source and Commercial software worlds (remember WordPerfect Office for Linux? Most people don't, which is a bit of a shame).

      One of the main advantages of Open Source, according to Richard Stallman, is actually that it protects against this problem. The theory is that if a piece of software were valuable enough to its users, but not valuable enough to the original author, that someone else could easily pick it up and run with it. Commercial software often doesn't have that advantage, as the source can die with the company.

      In practice, of course, we rarely see this. Either a software program is valuable or it is not. If it is, then whoever develops it will keep on developing it and making money at it, or they will sell it to some other company who also sees value in it.

      And, with the exception of software where the users are inherently technical (such as virtualization software, web servers, operating systems, development platforms), the people developing the software generally need guidance from people who use the software. This is really where many Open Source projects run into trouble - the developers can choose (and often DO choose) to lead the projects on their own, and really have no idea what is needed to be successful. In a closed-source/commercial environment, the project manager, software architect, and business analysts have far more control over the direction of the software, as they should.
      • Stallman's analogy is flawed

        if you are worried about the viability of a company or its ability to continue to support and develop a production, software escrow agreements can be made.

        It's a push.
        Your Non Advocate
      • Yes... yes it is...

        "Stallman's analogy is flawed"

        Stallman's theories are based on the premise that anyone can write software. Really, his theories are sound, so long as you take them in this context.

        The minute you start considering the real software development world where most programmers are paid to write software for other people who have better things to do, his theories kind of fall apart.
    • As compared to costly proprietary dead-ends or lockins?

      Are you sure the cost of being stuck on a proprietary vendor's solution is less? Or the cost of choosing the wrong proprietary vendor and having to re-do after that vendor's product has lost the race? With an open source solution you can at least continue with that solution and tweak it as *you* need to, even if others quit working on it. Often, the open source solutions have pluggable capabilities, so switching out for a different provider for one or more is possible...vs. the proprietary everything-from-us design.

      There are risks to all options; you are not making a case that open source risks hurt worse than proprietary ones.
      • Drinking Red Hat's kool-aid, are we?

        "Are you sure the cost of being stuck on a proprietary vendor's solution is less? Or the cost of choosing the wrong proprietary vendor and having to re-do after that vendor's product has lost the race? "

        Red Hat likes to trot the whole "Open Source = Zero Vendor Lock-In" myth (and obviously, many FOSS advocates actually believe this), but it simply isn't true. Any virtualization solution is going to lock you in to some degree. Having the source code available only mitigates the risk to the degree that the customer can write and understand the code themselves, should the vendor and/or community decide to abandon the software.

        Third party partnerships and backing by a solid, proven technical base, and good ol' profitability trumps availability of source code when it comes to mitigating lock-in risk.

        "Often, the open source solutions have pluggable capabilities, so switching out for a different provider for one or more is possible...vs. the proprietary everything-from-us design."

        Implying that proprietary vendors don't offer "pluggable capabilities" nor adhere to standards that can help mitigate lock-in. In the current context (VMware/Microsoft vs. Red Hat, Xen, KVM, etc) you are not only wrong, but you are wrong to the point that you look like a complete idiot trotting out that line. Come back when you've done some homework.
  • Seriously, Don't expect VMware to agree with RH on anything

    Red Hat are their 'worst nightmare'.
    DTS - Your Linux Advocate
    • If Red Hat are VMware's worst nightmare...

      ..VMware Execs must sleep pretty good at night.
      • Think again

        VMware made ESXi free because KVM is free.
        VMWare has redone thir entire pricing structure because Redhat had them beat on both up-front and TCO pricing.

        VMware is very mindful of everything Redhat does.
      • Re: Anothercanuck

        "VMware made ESXi free because KVM is free."

        ESXi is free? That's news to me. Probably news to the thousands of corporations each paying thousands of dollars per year in licensing fees to VMware, too.

        You mean free if you use it on one server. With only one processor. With 16GB or less of total virtualized RAM.

        In other words - free for home use and MAYBE the odd mom and pop small business. Certainly not free for use in a typical virtualized environment. Not by a long shot.
      • Windows Server Hyper-V

        I thought Vmware would more scared of Hyper-v 3.0 more then red hat personally.
      • @anothercanuck

        Some form of Vmware was free for nearly the entire history of VmWare. Remember GSX? That predates KVM.
        Your Non Advocate
      • Yes, ESXi is free

        @daftkey: ESXi is free to anyone, or any company, for any number of servers/cpus/ram. VSphere Enterprise and Enterprise+ are what cost big $s. But don't take my word for it, visit VMware's website.
        Without the enterprise version of VSphere, ESXi is only a hypervisor with basic management tools, no HA, no fault tolerance, meaning it is near useless for more than a few servers.

        @Viper589: HyperV is not a threat to VMware or KVM in the enterprise virtualization area as even the new, yet to be released, HyperV only supports 2 guest OSs (Win2003R2 and Win2008R2) with High Availiblity, and lacks fault tolerance. Not to mention HyperV is way behind in VM performance.
  • OpenStack and the Virtualization Picnic.

    VMware: "We're used to having our own private picnic, and we do pretty well. Lately we've been seeing some of the "other guys" walking in and stealing our food though, so we're thinking about just inviting them. We're open to sharing a bite or two of our cloud sandwich, as long as it keeps the competition away from the rest of our picnic basket."

    Red Hat: "We came to the virtualization picnic only to find that VMware had first dibs on the buffet. By the time we got there, Microsoft, Citrix, and a bunch of other free virtualization players already finished eating and there was nothing but a couple deviled eggs and a pickle left for us. Next time, we rent a bus to take everyone to the picnic, and we'd better be driving!"

    NASA: "We made a nice salad - I hope everyone likes it."

    Wall Street: "The food looks great at the picnic - especially the salad! I'm not going to buy tickets to eat though - salad gives me heartburn - but I definitely will recommend everyone else go for a bite!"

    RackSpace: "We cater virtualization picnics every year. VMware always has the biggest crowds with tons of food and can easily keep us in business themselves. However we could always use more business. If we can find a way to get all the other virtualization guys to have their picnic at the same time, then maybe we could almost have TWO big catering gigs!"
  • Simple virtualization

    For desktop virtualization, ThinServer XP or ThinServer Win7 does a pretty decent job :)
  • Commercialized product pricing is ridiculous

    The open source play is perfect for the orchestration component if you have the features you needed without a vendor lock. The consumer never sees this abstraction like for example a desktop running nix or windows.

    I just got done installing on a lab box using this tutorial. Pretty easy since I am an applications guy. Could get confusing without good directions like most big open projects. Documentation on the wiki is a little tough to go ground up with.