Red Hat Enterprise Virtualization (RHEV)

Disclaimer : Warix Technologies is one of the partner for Red Hat Enterprise Virtualization in Malaysia and I am a Fedora Ambassador. I have tested the late BETA of RHEV for a few days.

So Red Hat had already released their new and shiny Red Hat Enterprise Virtualization or RHEV in short, a few hours ago. This release, is one of the key milestone for RHEV after being officially announced in February 2009. So what this really means?

Virtualization is not something new, in fact its already nearly 50 years since IBM put their Virtualization solutions in their mainframe (and still is doing so on their System Z). But lately the surge of Virtualization hits new heights the last couple of years in the x86 market due to one factor, VMware.

Shortly after releasing their VMware workstation to the market in 1999 for the workstation and desktop market, VMware realized in order to be a big hit, it needs to play in the datacenter. So they start to focus their efforts to ensure their solution works reliably on the greatest of workloads and datacenters.

The Open Source Community picks up the fever soon after with the Xen Project in 2002. Xen the project and XenSource the company fuels the new technological goldmine and became for a while, the de facto standard. Citrix realized this and bought Xen for USD500 Million soon after.

Red Hat, Novell and other Open Source vendors jumps into the Xen fever and introduced it to the world circa 2006-2007 until now but hardly makes a dent into the VMware market either through lack of marketing, less features and I think most important of all IMHO – Manageability.

Around early 2007, enters KVM from Qumranet which turns Linux into a Hypervisor. But thats not all, they even build a user friendly web management interface that can manage big numbers of virtual machine, though mostly for Virtual Desktop Infrastructre (VDI) or desktop virtualization. Red Hat then acquired Qumranet and move the direction of the software to also manage the servers.

Enough of the history lesson, so basically RHEV have 4 components which are:-

  1. RHEV-M for Server (or Management Interface for Server) – This is basically the web-based GUI interface for RHEV solutions with ability to manage hundreds and thousands of server using its unique search-driven interface. This is the equivalent of Virtual Center of VMware.
  2. RHEV-M for Desktop (or Management Interface for Desktop) – Same with above but to manage your VDI Infrastructure. Also includes user portal for user to access their VDI. Release TBA.
  3. RHEV-H (RHEV Hypervisor Baremetal) – The baremetal Hypervisor based on the same code with RHEL 5.4 with less flexibility but easier installation for Non-Linux Admin / User. It can also be installed on USB disk, memory cards on CD.
  4. Red Hat Enterprise Linux (RHEL) 5.4 with KVM support – The vanilla RHEL that we know and love (or CentOS for some of you) with KVM and RHEV management support. This basically means you can hook-up your RHEL into the RHEV-M to manage your servers.

So what’s great about all this RHEV stuff?

  1. KVM – Yes that is single biggest thing in KVM that will be the game changer. Unlike Xen, the application compatibility betwen Linux in Baremetal and Linux in Virtual environment is gone as the 2 share the same code base. So whatever runs in baremetal, will run under KVM without any modification. This will ensure compatibility between binaries and ensure less headache to software vendors to create version specialized for VM environments.
  2. Manageability – Virt Manager to manage big numbers of VMs? With all due respect, NO. Enter RHEV-M. The one massive advantage that VMware had over other Open Source Virtualization vendors will be no more soonish. RHEV-M enables administrators to manage their Virtual infrastructures without knowing the command line that Linux is famous for. Its point and clickfest all over on their favourite Internet Explorer’s .. yes IE and that one will be talked about later :p
  3. Open Source – When talking about Open Source, there’s 2 great things. 1 is the cost factor and no 2 is the features factor. The offerings from Redhat is cheap compared to its competitors (USD499 for standard and USD7xx for premium per socket) and the features will keep on coming (libvirt, SRIOV, Libguestfs etc) when you subscribe to RHEV due to active community around KVM and its companion.
  4. Memory and Storage Overcommit – Thanks to QCOW or Thin Provisioning and KSM or memory page paging, its now possible to allocate more memory (RAM) to your server up to 150% and same applies to your storage (though maybe not 150%).
  5. Paravirt Drivers – RHEL 4.8 (and above) and RHEL 5.3 (and above) automatically accelerated with VirtIO Paravirt Drivers that will boost the performance of the VM to near native. For Windows, the paravirt drivers is WHQL certified and will be delivered through Windows Update. Cool rite? Thats what you get when you in bed with the devil!

What is NOT so great about it?

  1. M$ - The biggest issue about RHEV (and most controversial) is you have to run the management web interface on Windows Server. Yes its not a typo. And its also needs to have (take a breath) MS SQL, Windows Presentation Foundation, Active Directory, Windows POWERSHELL and also ONLY supports Internet Exploder.. opps I mean Explorer 6,7 and 8. And NO you cannot run it on Mono or Wine or Crossover for all the stuff above. But on the other hand, the potential customer might even be OK with it as they already have their Windows Infrastructure in their organization.
  2. Scripting – You can now do scripting to your RHEV, but only supports POWERSHELL!! no bash for you for now, though the VMware guys might be OK with it as VMware also supports powershell.
  3. I cant think of any more major roadblock, So just a placeholder if I remember any.

All in all, RHEV is a solid initial product, thanks to Redhat, Qumranet, Open Source Model and all the communities. This is what I call the power of participation. Although they are certain things that doesnt appeal for purist and Open Source community, I think Redhat have explored all the possibilities and thinks this the right way to go for now.

The new version (which will be RHEV 3, the current is RHEV 2.1) which targeted by middle of next year will runs on JBoss and Hibernate (which means you can hook it to almost all database) which should make everyone happy except VMware off course.

About these ads

21 thoughts on “Red Hat Enterprise Virtualization (RHEV)”

  1. Thanks for quite an interesting take on the subject. I’m the “purist” you mention… but all in all I think I was quite tempered.

    If Microsoft released Hyper-V and it’s management app was Linux based, how well do you think that would go over? :)

    I’m not saying I personally have a problem with it (I was shocked at first but I got over it)… but I think the “community” might.

  2. Scott Dowdle,

    You have a point but it should be clear that Red Hat didn’t develop the product from scratch. It was Qumranet’s decision before the Red Hat acquisition and the option was to throw away a lot of work and start over from scratch or ship the existing product with more features and go for a cross platform next release.

    Considering that the primary target would virtualising Windows desktops, having a management solution that runs on Windows is not a insane decision although bound to be controversial coming from Red Hat.

  3. One thing about Redhat always make sure they did before getting any product out is ensuring that it is open sourced or will be open sourced in the future. They also take the extra mile to make sure that they will not alienate their main source of the entire ecosystem, community and developers. The cross support agreement with Microsoft shows their commitment unlike their greener competitor.

    Specific for RHEV, only the management tools are not open sourced and the VDSM protocol (that interconnects virtual machine hosts and the management console) that is not open sourced for now. And they will open sourced within 6-8 months. Cross my finger for that :)

  4. You guys are preaching to the choir. I drank the Red Hat koolaid many years back and have not regretted it.

    I mention in my posting the same things you mention… that Qumranet made it and what the options where. So, I’m not really a “purist” but just trying to address what some in the community might be thinking… and trying to answer to it.

    For more on my Red Hat related thoughts see:

    Is Red Hat still relevant? You bet –

    http://www.montanalinux.org/redhat-relevent.html

  5. I don’t get how red hat is pushing on kvm this much while the project’s releases are getting less and less frequent. Are they holding back the good stuff?

  6. Hi Noname,

    I seriously remember its per server, but thanks for pointing out. will clarify the matter maybe tomorrow.

  7. im sorry.. but where does it state that red hat will release a linux based management console at all..? have they even made a statement about that..

    >”But on the other hand, the potential customer might even be OK with it as they already have their Windows Infrastructure in their organization”<

    really..and what about the potential customer that does not… those are extra cost that they conveniently do not add in their costing page… this is as bad as google toughting how much linux they use yet when an app comes out they only support windows…

    while i may not be a 100% purist, i think this is a big black eye in the face of the OSS community… yes vmware does require windows to manage their environment.. but vmware does not claim to be open source or even tought the use of the linux kernel.. but red hat.. COME-ON!!!

  8. Petem,

    Based on my information, the new RHEV-M for servers version 3 is already in the works based on JBoss and Hibernate and will be released around the 6-8 months timeline.

    Forget to mention that if the customers doesnt have Active Directory (AD) in house for now, the build in AD in Windows 2003 will be sufficient.

    And looking into the marketplace, for VM sometimes POC is around 4-6 months timeframe, by which time the JBoss interface will be ready when the customers moving into production mode.

    Just my 2 cents. (I dont work for Redhat)

  9. Just got off the phone with Red Hat. Give ‘em some time. RHEV is slated to be fully FLOSS in maybe a year and and a half. They’ve come through before with other previously closed-source products. I don’t see why they wouldn’t again. For example, Netscape Directory Server (now RH/Fedora Directory Server) is fully FLOSS.

    They’ve come through before, so if history is any indicator, they will again. It’s a lot of work to undo Microsoft-proprietary bits and make it cross-platform. :-)

    –SYG

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s