What Gartner can’t – or won’t – tell you about OpenStack Cloud

Gartner is a church of the IT industry. Those, who make large donations to “Gartner Church” get the blessings. Those who don’t are shunned and excommunicated.

On a related note, Gartner recently released its “beware of OpenStack” manifesto. Despite being written by Lydia Leong, whom I consider to be the most credible priest of the Gartner  Church, from my vantage point the report sounds unjustifiably biased towards some of Gartner’s best customers  and contains some misleading statements.

Here are some of the things in the report that illustrate my point:

    1. “OpenStack is dominated by commercial interests, as it is a business strategy for the vendors involved, not the effort of a community of altruistic individual contributors.”

      Yes, it is. So what? Nobody ever claimed otherwise.

      The key is that OpenStack is the only open source cloud community with meaningful involvement from many commercial interests, working collaboratively towards a single goal. VMWare is just VMWare, Eucalyptus is just Eucalyptus, CloudStack is just Citrix. OpenStack receives meaningful support from the entire application infrastructure ecosystem. (I explored the importance of this exact topic in an earlier post.)

    2. “OpenStack is not an open standard, as it does not provide for either interoperability or portability.”

      It is definitely open and, at this point in time, it is as close to a standard as it gets in the open source cloud ecosystem. Making the leap from interoperability challenges to “not an open standard” is unjustified.

    3. There is no significant difference in lock in for customers adopting OpenStack than those customers adopting proprietary CMPs.”

      If I want to deploy OpenStack, I go to openstack.org, deploy it using freely available recipes from Puppet or Chef and I have myself a cloud. I can do it with almost any hypervisor, use almost any flavor of Linux as a host OS and run it pretty much on any hardware. Whatever is missing, I can always build and add myself. Any new OpenStack features are always freely available. No matter how big my cloud grows, I never have to pay any money to anyone if I don’t want to.

      If I want a VMWare cloud, I have to call VMware and pay money to get their product. If I want to use Xen with VMWare vCloud – I can’t. As VMware releases new features, I have to pay more money to buy them.

      What am I missing? Perhaps I am not clear about the meaning of ‘lock-in’?

    4. “Consider proprietary CMPs as well as free and commercial distributions of OpenStack and other open-source CMPs, particularly Eucalyptus (which has officially licensed the AWS API from Amazon and is relatively easy to deploy) and CloudStack (which is a common choice for service providers and others wanting to run an AWS-compatible cloud at scale).”

      Let me say a few things here about Eucalyptus’ API licensing deal with Amazon. It is nothing more than a press release. API’s don’t need to be licensed; they’ve already been ruled to not be copyrightable on at least one occasion. Beyond supporting Eucalyptus with a quote for their press release, Amazon doesn’t care about them. Eucalyptus would align with OpenStack marketing hype if it could; but it can’t. So it aligns with AWS marketing hype. That is it, cut-and-dried.

      Don’t consider CloudStack. It will soon die. True, it is getting some traction in the enterprise because Citrix is pushing it, using its enterprise sales muscle–plus it has a pretty UI. But the truth is that OpenStack has already surpassed CloudStack feature-wise. More importantly – hype or no hype – RedHat, Canonical and SUSE chose to align with OpenStack over CloudStack. Because of that, CloudStack only runs on Ubuntu 10.04 host OS, which is 3 years old and doesn’t have driver support for some of the newer hardware. As time passes, this will only get worse.

    5. “The open community process, in combination with closed-room discussions between the sponsors, sometimes results in intense infighting between the participants in the community, particularly as Rackspace works to establish the OpenStack foundation.”

      This is just not true. There have been only a few board meetings so far; all – very productive with no infighting. I can attest to that as someone, who personally participated in ALL meetings.

      The reference to the executive board sessions as “closed-room discussions” here is clearly meant to carry a negative connotation. I am unclear how these executive board sessions are a bad thing for the community. Please explain to me in the comments below if you feel otherwise.

  1. “The projected time frame for OpenStack stability is constantly moving… Commercial software normally reaches stability by this age and the projected time frame is exceptionally long, even for complex open-source projects.” 

    Stability is one big gray area and is a function of adoption, not time. Everybody knows that enterprise software sucks – commercial or open source. I can just as credibly make a claim that VMWare, Microsoft and Citrix are not stable and never will be.

    OpenStack is the biggest open source cloud community, which means that more people are tinkering with it than with any other open source cloud product out there. This, in my opinion, means that OpenStack wins the case for stability. At Mirantis, we have deployed plenty of stable OpenStack clouds. I openly invite anyone at Gartner to speak with our customers.

  2. “OpenStack will face a very difficult battle against VMware, which already has a significant installed base, channel to market and suite of products that provides far more than resource management.”

    I am hoping that VMWare will help OpenStack in that battle, now that it joined the foundation and pledged to support and empower OpenStack software =).

    VMWare is strong in the legacy, datacenter automation market. I wrote about it as well. OpenStack is competing more with AWS, not VMWare in the new, disruptive “open-cloud market.” It is true that, for now, enterprises fail to see the difference between VMWare and OpenStack. Longer term – it will change. Just like Blockbuster’s foothold in the retail video-rental business didn’t help it win against the on-demand video, VMWare’s foothold with enterprise virtualization has questionable value in the context of this market transformation. Its installed base and channel are, arguably, its liability, not its asset. Cloud spin-off could have been a good idea.

  3. “To maximize deployment flexibility and interoperability to enable the potential for multi-vendor substitution, try to choose CMP solutions that allow the layers of the service (access management, service management, service optimization, resource management and the underlying resources), to be logically independent of one another.”

    This, in fact, is exactly the reason to consider OpenStack. Unlike with VMware or Microsoft, OpenStack is designed as a series of loosely coupled components that are easy to integrate with a variety of third party solutions and hardware platforms. The only reason why it doesn’t make sense to use OpenStack with commercial platforms like VMWare is because VMware’s hypervisor is only designed to work with VMware’s suite of products. “Maximizing interoperability…for multi-vendor substitution,” as suggested in this report is only possible with OpenStack and not commercial offerings or vendor centric, open source solutions.

The overarching theme of Gartner report is “don’t be blinded by OpenStack marketing hype.” I agree; do your homework and consider various options. It is true that OpenStack gets a lot of hype.  But don’t discount OpenStack hype. In open source, hype translates to real value. It is because of that hype that OpenStack is now supported by everyone in the application infrastructure ecosystem. And is because of that support, it is a truly great community that is well on its way to become the Linux of the cloud.

33 responses to “What Gartner can’t – or won’t – tell you about OpenStack Cloud

  1. Do you work for Rackspace?
    This article is terrible
    The point of the Gartner report is not about focusing on OS vs VMWARE but instead you use VMWARE as a scape goat to make Openstack look better then it actually is
    shame on you
    ITs still a project with only 2 customers (partly in production) – HP and Rackspace. Thats the truth and we need to be honest with ourselves and customers and not be blind sighted by the amount of people joining a community. HP, IBM, CSC, DELL, RED HAT, Cisco all have competing interests! good luck with making all of that work! I like your positive spin on it but the OS governance model sux…which is why I have committed so many developers on Eucalyptus

  2. Straight shooter, huh? My kinda guy. I appreciate your candid feedback and, naturally, don’t expect much love from a Eucalyptus fan.

    I believe that your comment exemplifies the reason I wrote this post. There are a lot of people like you that read the Gartner report and then re-post all over the web that OpenStack only has RackSpace and HP as customers… this is just not true. How about Webex that’s presenting at the upcoming summit? How about Gap? Dreamhost?

    Re: governance model – I would suggest to think about the following…how many competing interests are sponsoring the Linux Foundation and how much of a hindrance to Linux’s success has it been so far?

  3. I’m respecting Gartner’s rules for not debating published research in a social-media forum, but I did want to make one comment for Tedd Harrison:

    The report doesn’t assert that OpenStack has no customers, and certainly does not intend to imply that. It states, “At present, OpenStack early adopters are organizations with strong technical skills and a high risk tolerance. They are typically service providers, financial services companies, or research institutions.” We’ve had discussions with numerous customers and prospects outside of those verticals as well.

    1. Lydia, thank you for clarifying.

      No disagreement re: “strong requirements for technical skills and high risk tolerance.” I’d like to add that that same is true for any new, particularly open source, offering.

  4. It does not matter what any of you think, VMware and Amazon have the market wrapped up. Eucalyptus Systems is smart working with Amazon in whatever way that it does. Best to ask Marten what the relationship is between the two companies, most of you don’t really know. It will be so long before OpenStack is ready that we will be onto our next big project.

    1. Way to stay positive Bill. If I had a dollar for every time someone made a market prediction I’d be a billionaire by now.

      VMWare just laid off 900 people. If they have the market so ‘wrapped up’, what’s up with that?

      OpenStack is production ready today. RackSpace has it in production. HP has it in production. Dell is putting it in production in Q4. It wasn’t so long ago that AWS was still in beta. Things take time to bake properly.

      We’ve gotten an awful lot of cooks in the kitchen now working on OpenStack. I really don’t see that happening with Eucalyptus. Ever bother to look at the commit logs? You have 10X the activity on OS compared to Eucalyptus.

      Do the math man.

  5. Actually I think the best thing going for OpenStack is it’s extensible API. The problem with Eucalyptus and cloudstack implementing Amazon AWS API is they are limited to Amazon AWS features. HP Cloud and Rackspace use the Metadata fields to add their own features. Thus the OpenStack Cloud could become the open API cloud standard because of this extensibility. This enables companies to innovate in the cloud area while still adopting a standard.

    1. Neill – very good point. I agree completely. When considering API compatibility people too often forget that API is there to expose the underlying features. Obsessing over support for AWS API also implies submitting yourself to AWS feature roadmap and stifling independent community innovation.

      1. Apahce (Incubating) CloudStack has its own open API, and alsoo has an AWS API compatible module. It has a rich API and a way to plug in and extend the api to your liking. It also runs on ubuntu 12.04 and future builds. I think their community addressed that well.

  6. Boris,

    Awesome post. Its a daily struggle to rationalize the differences and value of OpenStack versus solutions like VMWare and Citrix, especially when enterprise directors and managers get fed this kind of propaganda from “independent sources” like Gartner.
    It took the IT world years to finally relent that Microsoft was not the only OS player in the game and time and economics were Linux’ best friend; I suppose the same will hold true for OpenStack.
    BTW, ATT has many tenants in production across multiple DCs.

    regards

  7. First off, the point about cloudstack running only on Ubuntu 10.04 is not accurate. The features description is also not accurate – CS has many features that OS lacks, especially on the networking side.

    I find the statement that Cloudstack will die soon a bit short sighted, if not dangerous. The competition can and should never be underestimated or disrespected. Products don’t really die out “soon”, especially when backed up by good support, coherent points of contact for customers, and a long term roadmap that backs major investments by customers. CS has all of these and is delivering consistently on new features, and it has a healthy base of non-enterprise customers. Plus, it’s open source, incubating with Apache, and has one of the fastest growing opensource communities out there.

    Not to forget that CS works with KVM, XS, ESX, HyperV, OVM – the lockin-free advantage isn’t a monopoly of Openstack alone. Configuring Network as a service is a superior experience in Cloudstack. It has extremely good support of external devices across vendors, and its feature set within its virtual routing is rich. Indeed, Openstack is extremely customizable – but cloudstack is too, and the customization needs of specific customers is well taken care of in its design.

    From among AWS, Eucalyptus, OS, Azure, Vmware and CS, I actually find Citrix and CS altruistic. There is nothing stopping others from picking up Apache Cloudstack and commercializing it. There aren’t vested interests that try to push around CS design in a particular direction or towards a particular hypervisor. Features are uniform across hypervisors, and equally rich with focus on both both enterprise and non-enterprise customers.

    When I interact with people who’re just starting off on the cloud, I find them struggling with getting Openstack up and running, not owing to it’s complexity (which cloudstack also has in ample measure), but because it throws too many modules and terms at the end user, and is hard to put together. On the other hand, people tend to pick up cloudstack much quicker – it’s not a simple matter of better GUI that makes this happen. It’s more than that. Cloudstack is elegant. I don’t expect everyone to feel the same way, but it comes across as the unofficial consensus. So, go through its design and its code and make your own assessment. You will know what I’m talking about. Same goes with documentation – Openstack has excellent documentation for individual modules, but putting them all together is hard for novices. Cloudstack documentation focuses on deployment, which is what businesses really look for, and delegates details of individual functionalities to its cloudstack wiki pages.

    Next, actual product development – as a developer who has worked on both OS and CS, I feel that python based frameworks with or without IDEs are great for hacking, but development with Java based ones with IDEs is much more rapid and robust. This may be a personal quirk/preference, but I’m noting it anyway since this is a major difference between CS and OS.

    Don’t get me wrong here – OS’s design is really good, and I love Openstack. Nobody doubts the fact that it’s a great product with some shortcomings to address, and definitely a few strengths that other products lack. Not to forget the massive support it has received from major companies. But that very fact can slow it down. So I’m not going to rule out or sling mud on other products, especially a good one like cloudstack.

    Disclaimer: I work at Citrix, but the above views are personal.

    1. “Configuring Network as a service is a superior experience in Cloudstack.”
      Sorry but No, I am a CloudStack user, and I can say this, configuring networking is by no means easy for alot due to the model chosen and lock in required to comply with your model. not allowing multiple nics per VM in basic mode and forcing the use of VLANs is painful to alot of us. I still completely agree with the rest of what you said.

      1. I am also a Cloudstack user, and I find the networking functionality of Cloudstack to be the best of any cloud platform I’ve used including AWS, Openstack, Onapp, and Eucalyptus. The use of VLANs is THE RIGHT WAY to do networking between disparate computing user accounts in any network, virtualized or not. On a private cloud, use basic networking, it just works. On a public cloud, at a service provider, advanced networking is the only way to go. I’m not throwing mud, but if you think “forcing the use of VLANs is painful”, then maybe you should not be working in a cloud environment. Unless you’ve used Cloudstack in > 3.0 release, please don’t judge the usability and stability of Cloudstack. The OP’s opinion that Cloudstack is plagued by Vendor lock-in is 100% FUD. In fact, Cloudstack supports more platforms, both virtualization and network edge appliances like F5, Netscaler, and Juniper. It also supports Xenserver, XCP, VMware, KVM, Oracle soon, and bare metal. Last time I counted, that was more than any other platform. I don’t work for any of the companies metioned by Gartner, Boris, or you, and I took a neutral approach when selecting a CMP, and Cloudstack came out ahead by a wide margin purely based on the networking capabilities and hypervisor support.

    2. Thanks for a detailed and thorough commentary.

      My conclusions on CloudStack are not based on things I read or heard others say. We’ve been working with CloudStack in our lab and have also recently published an OpenStack comparison – https://www.mirantis.com/blog/an-openstack-guy-takes-cloudstack-for-a-test-drive/.

      And, again…no, it doesn’t work on 12.04. Peter Ulander was just as amazed to learn this as I was, judging from his recent twitter exchange with Kevin Jackson =)):

      Ulander: @MirantisIT CloudStack supports current Linux distros http://t.co/iWEMGpyA No need to be deceitful in your marketing, reflects poorly on you
      Itarchitectkev: @ulander That page says Ubuntu 10.04 – that’s 2 1/2 year old distro – I’ve tried CloudStack on 12.04 and it fails /cc @MirantisIT
      Ulander: @itarchitectkev @mirantisit fair statement, that said we build for our customers environments. Your employer seemed to be fine with that.

  8. Uhhhh your an idiot! cloudstack is far more then just a pretty gui. and your completely WRONG on this point where you claim “CloudStack only runs on Ubuntu 10.04 host OS, which is 3 years old and doesn’t have driver support for some of the newer hardware. As time passes, this will only get worse.” FUD from Gartner as usual, but from you? Even more FUD in reply. Id expect more from the likes of you. Lets be fair.

      1. It simply doesnt just run on Ubuntu, it runs on CentOS also, note in your own link where he states “I did not try Red Hat or Centos with CloudStack but found similar complaints here”. He also plainly states “Eventually, I gave in and downgraded to Ubuntu 10.04 (almost three years old) Why not Ubuntu 11.10 then I ask?? So as you can see, your article is A) Biased, slanted and not very concise, B) Not very well researched. Dont get me wrong, anything Gartner has to espose is evil and misguided. But from an open source and a projects perspective, you spread their FUD even further, by wrongly discrediting other projects, when your research is either invalid, or incomplete. Lets to say even the level of expertise the user that attempted to make was even a viable one, with a conscientious effort even. Dont get me wrong, Im in whole favor of any effort to produce a great environment, but lets be fair about our reviews. And I would also note that the bug report looks to me like its in reference to using 12.04 as a vm instance, not as a CloudStack Management server. Matter of fact, Im going to even validate what the poster claims, more from curiosity.

        1. We are talking about using Ubuntu 12.04 as a host OS here. Do try it yourself and let me know what you find out.

          My point is that it’s a bad symptom. Cannonical, RedHat and SUSE all opted to support OpenStack, not CloudStack. Long term that means no support for CloudStack from the Linux community. It is now solely on the shoulders of Citrix to make sure CloudStack works with newer Linux releases, and eventually Linux kernel itself – not a good thing.

          Not being able to use 12.04 as a host already today is symptom of a trend that, when projected into the future, makes me think twice about putting money behind CloudStack.

          1. No, you stated “CloudStack only runs on Ubuntu 10.04 host OS, which is 3 years old and doesn’t have driver support for some of the newer hardware. As time passes, this will only get worse.” This is completely untrue, dont try to justify it by back pedalling and change the inference to your statement. Be a man and admit you simply didnt do your homework and guilded the FACTS. Cloudstack runs on Ubuntu/Redhat/CentOS, not just Ubuntu 10.04, as you stated in your own rhetorical fud.

        2. For the record, I run CS 3.x on CentOs 6.2 with MariaDB for the database. CS is java, it runs on pretty much any OS that runs java. The article is choking with FUD.

  9. I also think it’s silly to say that one will “WIN” and the other will die. These are large markets that are almost always at least duopolies. IT vendors are supporting both and developing both platforms. This guy summed up my thoughts pretty well: http://learnitwithme.com/?p=349

    I think in general OpenStack is more flexible, being a bunch of pieces to put together how you see fit, but as such requires more expertise. From the customer perspective it will reduce the ability for customers to go betweeen OpenStack vendors since each will have different solutions/components.

  10. Gartner doesn’t write manifests. If you need an reminder of what a manifest is, I’d suggest reviewing the Unabomber’s.

    The recommendation is spot on for CIOs trying to figure out what to do CMP-wise. I suspect you have very little large-enterprise IT experience. Hype is why the college grad developer is trying to convince management to use Node.js for the financial system rewrite. The CIO has 500 developers worldwide who’ve used VMWare on their desktop for the last few years. He likes his VMWare sales guy, the company has tons of VMWare experience, it’s designed to work with his $1mil EMC install, and he doesn’t want to explain to the CEO that the datacenter’s down because of an open source bug. He wants to say it’s down because of VMWare, but they’re giving a credit for the downtime.

    She even tells those who use OS to join the community and submit patches! Lydia, could you tell your customers to do the same for CloudStack? 🙂

    Would love to hear which Gartner customer you think requested this research. From my experience, analyst puff pieces are usually pro-customer, not anti-competitor.

    The problem with an open source product being dominated by commercial interests (Like Mirantis’) is the community will be left with a product lacking key features. I experienced this personally when I attempted to use OS about a year ago, and was unable to find a hosting billing package that supported OpenStack. Rackspace contributed the CMS, but didn’t provide the ability to track usage at that point. I haven’t talked to Rackspace, but to me it seems safe to guess this was done for competitive reasons.

    “Don’t consider CloudStack. It will soon die.” Ya know…I could say so many things to this, but what I’ve done instead is put a note on my calendar for one year from now to follow up then. So, we’ll continue this topic then. 🙂

    AWS provides a service that “just works.” OpenStack (or CloudStack) is a cloud management platform. They’re just not comparable.

    It’s too bad OpenStack doesn’t work with commercial products like VmWare – you should fix that. Feel free to look at the CloudStack code. 🙂

    Regarding hype – just remember, for Justin Bieber fans, screaming girls translates to good music.

    Disclaimer: I’m a committer and PPMC member of the Apache incubated CloudStack project. These views are my own and not on behalf of Apache CloudStack community.

  11. That is the kind of comments I really like to see here.
    I’m a newbie in Cloud, with a lot of exp in networks and TI, but I think OpenStack has a major success because in less than 2 years go from a project from companies to a consortium an becoming in a “pretty lady” in the cloud market.
    Keep in mind, almost all the partners are the same as competitors. That sound really really good!.
    I’m really impressed about this article. And by the way, In my country the people knows that Gartner is a reference to buy, design and deploy technologies…nobody has a doubt about this.
    In my experience, AWS has a major business but as all my old customers always complaint to me : “I don’t like this model because is almost proprietary and not so flexible”. That’s a fact.
    I don’t know if the other guys in open source solutions has the same level of “partners”, but at the end of the day, 2 years of OpenStack consortium, deploying a tool box to provide private cloud, is a success.
    Best regards, and respect.
    Marco Bravo.

  12. I would really appreciate if someone could provide me with a link or any material which has a complete comparison of the entire feature set of Cloudstack and Openstack…. I think Cloudstack 4 is going to take over the markets considering the strong support of the Apache foundation… At the same time I agree that Openstack too has come up with really superb features such as grizzly, OpenstackSwift etc., which cloudstack doesnt have… I am really confused on what platform should I go with… Can someone guide me with a feature comparison list….

Leave a Reply

Your email address will not be published. Required fields are marked *

NEWS VIA EMAIL

Recommendations

Archive

Live Demo
Hour with an OpenStack Expert
Wednesday
11am PST