Initial Reaction: Red Hat releases the SPICE protocol
Red Hat actually opened up the SPICE protocol yesterday during their Virtual Experience 2009 event. Somehow I missed that. Have a look at the press release if you are interested... as well as their site to house the new open source project - spice-space.org.
This seems to have caused some buzz in certain corners of the virtualization websphere (does anyone still use that word?) but so far no one has said what it could mean for us run-of-the-mill Linux users looking for a good, fast remote desktop protocol. I asked a couple of questions on the fedora-virt mailing list and received some informative replies. I haven't had a chance to actually check out the SPICE website in depth yet though. For those wanting technical documentation, I've attached their spice_for_newbies and spice_user_manual PDFs.
I did want to share with you my response to one of the articles about the SPICE release that I ran across. You can go there and see the context this came from but it is pretty self explanatory.
- - - - -
I'm glad you say that the open source community can't do it... and I hope more people hop on your bandwagon... because that'll just help ensure they can. :)
Regarding the open source development model, there are a number of benefits to it. It isn't all about free of cost. Big organizations prefer software they can have a say in. As an example, Google recently started a project that was a fork of FreeNX because they appear to want an open remote desktop protocol. I assume Google will jump right on this... but they will not likely dominate it... they would just be one of hundreds of companies that will contribute.
Will SPICE move out of KVM? Absolutely. It will become a general purpose remote desktop protocol used on both physical machines and virtual machines. I don't much care if all of the big closed products choose it or not. There are tens of millions of Linux users who want it and who will use it. Then there are the FreeBSD folks and the other free OSes that exist who will want it too. Linux has been missing a good open RD protocol... although I do realize that all of the commercial ones can be licensed on Linux.
I predict that SPICE will take some time to move outside of KVM. From what Brian said in the article it appears to me that there is a virtual driver needed to run SPICE... OF COURSE a driver can be written for physical machines too... assuming "driver" is the most appropriate word. In fact, detaching it from KVM is one of their stated goals.
Once SPICE becomes general purpose, assuming it performs even slightly better than VNC (the only thing needed for it to be successful on most open systems), it will become like OpenSSH has become... the de facto standard used by tens of millions of users of open systems.
Getting back to the big virt companies, who says they have to support one and only one protocol? If the code for SPICE is out there and gets refined by the open source community (which is often many paid people at different big companies who want an open protocol so they can stop paying for it and so they can adapt it to their particular needs) and adopted by tens of millions of users... the various virt companies would be foolish not to add it as an option.
Right now no one has stepped forward and picked up the ball but it was just released yesterday. Give it a few months.
So far as whether or not Red Hat Enterprise Virtualization will have a shot at being one of the major players goes... that is still up in the air. As is probably obvious, I'm a Red Hat fan so I certainly hope they will. Red Hat has been very successful with RHEL and a few other of their products and services... and their existing customers may well be interested in expanding their relationship with Red Hat especially given RHEV's low pricing strategy and simple licensing (no addons to buy).
Whether RHEV is a quality product that can compete with the big-boy feature sets now or one that will take time to mature remains to be seen. So far as the check marks go, they seem to be doing well but they need to prove themselves. I realize that often the better products do not win... and often lower quality products do better for often unrelated reasons... but virtualization is still young and there is plenty of room for the market to grow over the next 5-10 years. While VMware and Microsoft are market leaders in one or more markets that doesn't mean that they can't fall behind, screw up somehow, or butt heads with some actual competition.
Microsoft has an advantage with Hyper-V because it also controls a family of OSes that people want to virtualize and they can offer free OS licenses for the virtual machines (implemented in various ways) if they feel they have to. Perhaps Red Hat can leverage their position in the Linux marketplace into the virtual machine space. I would be surprised however if Red Hat ever did anything underhanded like so many others have.
One thing that no one has mentioned is that Red Hat believes in the Open Source model and they will eventually open up all of the code to the currently closed portions of RHEV. Right now the barrier is the fact that the current RHEV management application is based on Windows Server 2003 and a number of Microsoft technologies. Red Hat is in the process of building the second gen management product based on Java/Jboss to make it multi-platform. Once that is done and out they will eventually open up the code for that. How do I know this? Because they have done so with everything else they have done themselves and that they have bought from others. That's how they operate. They believe by opening it up, they can leverage their customers and the open source community to make the team bigger... and adapt the product in more and new directions that make it better... better than they could have made it by themselves. Red Hat sells support and services, not really bits... and they don't lose much of their support business when they open their stuff up. By opening their products they actually increase their customer base.
Regarding the lack of support for open source products and protocols... that is an opportunity. If you aren't happy with the fact that open source product X doesn't have anyone offering commercial support or that the existing support offerings are poor, someone else (even you) can change that and start a new support service. What it takes is massive adoption of product X to make the opportunity into a lucrative no-brainer. IBM has shifted a certain portion of their business to become a support company for open source... and they are a big lumbering company that takes forever to change direction. Imagine what a flexible, well run, smaller company could do.