So, want to manage a cloud with open-source software?

So, want to manage a cloud with open-source software?

Summary: Then, Convirture deserves your attention.


The sad, bad secret about cloud-computing is that setting up a cloud and running an application on any of them is relatively easy, managing it though, that's another thing entirely. That's where companies like Convirture come in.

Convirture's open-source project is ConVirt Open Source. With it you can manage both Xen and KVM (Kernel-based Virtual Machine)-based virtual machines (VM) and clouds. Specifically, with ConVirt, you can create and provision "gold" images, diagnose performance problems, and balance load across the datacenter, all from a Web-based interface and with consistent feature set across open-source virtualization platforms.

I've played with it some, and it works well. I especially liked that I could drag and drop live virtual machines between servers without interrupting their operations. This GPLed program allows you to re-allocate physical resources to virtual machines and balance the workload without incurring any downtime.

If you want to do more with it, Convirture has just announced a commercial product, ConVirt Enterprise Cloud. This, the company states, lets you manage private clouds, hybrid clouds and traditional virtualized infrastructure from a single Web console. With it you manage how datacenter resources are allocated between traditional enterprise applications, private clouds and public cloud resources such as Amazon EC2.

"Enterprise clouds are the natural next step in the evolution of the virtualized datacenter and IT staff should be able to manage cloud-based computing resources right along with existing virtual machines," said Arsalan Farooq, founder and CEO of Convirture in a statement. "Our customers are not interested in a 'rip-and-replace' approach, which leads to the separation of the infrastructure used to deliver cloud services from the rest of the virtualized environment. With ConVirt Enterprise Cloud, IT staff can build out a virtualized data center in the most flexible and cost-effective way. By managing traditional virtualized workloads side-by-side with cloud operations, ConVirt ensures a holistic view of the entire datacenter and allows IT staff to react quickly to changing business priorities."

From what I can see of it, this looks quite promising. I really like that it can give me management support for my local cloud, a remote datacenter private cloud, and public clouds like Amazon's offering. Speaking as someone who does a lot of networking work, I also like that I can choose to put a client on an existing corporate network, pre-configure Virtual LANs (VLANs) for them, or allow the cloud to create its own VLANs within a specified address range.

It also doesn't hurt that ConVirt Enterprise Cloud supports third-party private cloud infrastructure platforms such as Eucalyptus and OpenStack, so you can a centralized view of your virtualized datacenter. Since, at this point, there's no telling who's going to be the dominant cloud architecture, I like this flexibility.

So, in short, if you want to run a small cloud of Xen and KVM VMs, give ConVirt Open Source a try. If it works for you, and you need more, check out their commercial offerings. I think you might just like both.

Related Stories:

Amazon debuts ElastiCache; in-memory cache in the cloud

Is now the time for KVM?

Red Hat declares war against VMware on cloud front

Ubuntu: The desktop Linux with the cloud inside

Topics: Data Centers, CXO, Cloud, Emerging Tech, Hardware, Software, Storage, 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
  • RE: So, want to manage a cloud with open-source software?

    Steven, Great article.
  • Ditto. Nice write up.

    Waves Open Source Flag!
    Dietrich T. Schmitz *Your
  • Just a couple issues

    OK, first must have the resources to look at the cloud, and most companies are so tight on resources it won't happen any time soon.

    second, you must have the infrastructure to run the cloud, a tall order for companies with "iffy" internet access

    third you must want the cloud, and I don't see any real compelling reason to go to the cloud.

    fourth you must want to run your cloud on Open Source

    Fifth you must choose between two open source vendors, sounds like the good old MS lock in to me.

    As usual Steven waves the flag, but there's nothing backing it up. All hype.
    • Try reading the article, (or learn to read)....


      Yo state a couple (2) of issues but name 5 of them (although not one of them is valid!).

      "It also doesn?t hurt that ConVirt Enterprise Cloud supports third-party private cloud infrastructure platforms such as Eucalyptus and OpenStack, "

      And it states that 3rd party clouds are also supported. Where does it say only 2?

      Troll much????
      linux for me
      • You should get an editor

        @linux for me
        Ah, yes, when you disagree, misspell. "Yo state", did you mean You State?

        As for none valid, you might want to get out of your parents basement and try to get the resources to pull off a project like this. Good luck. Resources are tight and the technology is unproven. That translates into high risk.

        As for "Couple" - technically it does mean two, however, in real conversation it can mean more than two. Again, you ought to try basic communications. If that is beyond your capabilities, you really are better off in the basement.

        Now, run along and try to convince your employer (if you have one) to do the cloud thing, and use the products from the article. You might find reality crashing down, just like Steven does from time to time.

        I think I'll go out and enjoy the reality of life, you know, in the sunshine!
    • RE: So, want to manage a cloud with open-source software?

      @Cynical99 ...
      After reading your list, I really only see 3 questions there:
      1) Do you really want a cloud-centric solution?
      2) Do you have the resources (people, infrastructure, etc) to have one?
      3) Which vendor do you go with?

      (Actually, now that I think of it, that last one could easily be part of #2 ... but i'm willing to call it separate.)

      Obviously, those are all good questions that any CIO/IT Director should ponder, if only to make sure they aren't missing a good opportunity for efficiency, cost-savings, reliability improvements, etc.

      Which leads me to wonder: What exactly was your point by raising your questions? You seem to not like cloud computing, so why not state that at the start then explain why, instead of just implying that those who are interested in cloud solutions are wrong?

      I may or may not need a cloud solution where I am now, but it's worth looking at this offering, if only from a due-diligence and educational perspective, because who knows where I'll be in 2 years, and what my needs might be then?
      • perhaps it's because -

        The company I work for has already looked at clouds, determined that no vendor out there has a clue, the technology is too immature and pretty much useless in remote locations. Not to mention security holes you can drive a Mac truck through.

        Nice try though. Perhaps the clueless will jump on the early adopter bandwagon and after a few security breaches, massive outages, and the rest, they'll decide that clouds just aren't worth the hassle.

        If five or so years, perhaps the technology will mature thanks to the early adopters that took all the hits along the way, but somehow going back to the old mainframe days, thin client, and every other term used for central computing in the past 30 years won't happen without complete revolt from the user community.

        Good luck pushing that one.
    • Err..


      Interesting points you've thrown out there, but I find some of them wanting.

      1. Infrastructure: If you are a business that is indeed talking about a cloud solution, you would have to certainly scale. If you're an SMB with an insultingly low income, I sincerely doubt you would consider it on the scale you seemed to be hinting at. Either way, an SMB would usually have a minimum of a 20Mbps pipe (or OC-1, depending on the income- but in that case, my point is moot). Were that the case, you would have far more than enough to deploy a clustered and cached cloud solution without eating up as so much as 8Mbps for, oh, say 60 people? And that is a very generous ballpark figure.

      2. Resources to look at the cloud? Really? That's a rather silly gesture. It's a hard cold fact that you need resources for anyone to look into anything. Period. Don't get Derpington to do it, give it to one of the seniors that doesn't talk out his ass and you should be fine...

      3. I don't really see any real compelling reason to go to the cloud either- I definitely agree with you there. Distributed computing, I feel, has always worked better when it was kept within the boundaries of the company (for a plethora of reasons that I really don't care to list- but will if anyone wishes it so). OS Agnostic? That might be the only argument worth a damn I can conjure up right now, and even then it is rather flat and one-dimensional (since a proper setup is hetero anyways).

      4. I'm really not understanding what you're getting at here. In the aforementioned scenario, of some smaller SMB with a low profit margin (for schlitz and giggles, let's stretch this out a bit) it is generally more practical to go with a FLOSS solution due to licensing fees. If you've enough competent people around, management and any potential overhead arguments become irrelevant (and in this case, I would sincerely argue.... But that's outside of the scope of this response).

      5. Two? Where have you been?

      I'm currently overseeing a large VM deployment that is managed by Convirture EE. It's quite good software, although the js on the frontend can get a little sluggish when you swamp the server (to be expected, mind you). It's quite a good offering in general. Then open source solution does not fall from the tree either- quite solid, same js issues however.

      Aside for your (oh the coincidence) ad hominem remarks with regards to the grammatically/spelling-ly (lawl) challenged, it seems you're blowing a lot of hot air.

      Yes, early adopters will get burnt. Yes, a lot of vendors don't know wtf is going on- but not all. Security issues- only if you leave the task to derpington (And no, I'm not talking about something stupid like 'security by obscurity' efforts- talking about a proper setup to deploy secure cloud computing with minimal resources/time invested). The stronger the backend and overlaying technologies in use, the more secure your cloud topology is.

      One quick sidestep: Once again, I agree that the 'need' for Cloud is a bit pushed with buzz marketing (There's a very special place in hell...)- but it's not to say it is completely without merits.

      People say 'cloud' and they completely go by whatever is thrown at them. What idiocy (not by choice, but by ignorance led by flashy marketing and the stolen focus from more obscure projects).

      So what is left over? What about WASTE- sitting on it's throne of obscurity (ofcourse, studded with the finest diamonds, for it is brilliant)? Even going outside of the most commonly seen 'Cloud' solutions, there is so much bloody potential and software out there that could lend to the task that it is ridiculous. You just have to know what you're doing, and why.
  • RE: So, want to manage a cloud with open-source software?

    no one wants the 'cloud', it has been around since the 80's, and it has never gotten anywhere, for some very simple reasons.

    It is not possible to make you internet connection faster, or more reliable, or more secure than you can make your internal business network.

    2) you do not have to pay for the bandwidth of your internal network once it has been established.

    The weakest link in the chain is the link that will fail, if you make you chain much longer with many more links, you have many more ways for your system to fail.

    The internet will never be faster than you're internal hard drive.
    • RE: So, want to manage a cloud with open-source software?


      Interesting. Some good arguments you bring to the table.

      You're right. Your WAN pipe will not cure everything, and complexity can add to troubleshooting woes in the long run... But a proper OSPF/MPLS/etc configuration with redundancy, HSRP (VSRP preferred for obvious reasons) would definitely help.

      Some MPLS Traffic engineering is all you really need, and a competent network admin/engineer. Since Layer 3/4 is primarily your concern, L7 should be untouched/transparent. The only problem might be with TCP sessions dropping in the event of a switch-over (which also can be circumvented with some crazy mechanics like session caching and layovers).

      Your internal hard drive is indeed what makes or breaks something- however, this can be properly circumvented if you're worried about speed.

      RAID 5, 10, etc, with btrfs for everything outside of databases usually does nicely. You can fill the pipe with about 100 MB/s, at an absolutely low minimum (and that is damned low, for this setup, you should be expecting 200MBps+ with 15k rpm hard drives in the array).

      The idea of 'cloud' computing is interesting, but ultimately silly for being so prominent in the headlines as much as it is.
  • RE: So, want to manage a cloud with open-source software?

    It is not possible to make you internet connection faster, or more reliable, or more secure than you can make your internal business network.