One of the most popular discussions in the blogosphere is the topic of how Linux is posed to start taking market share from Microsoft in the battle of the desktops. What I don't see being discussed too much anymore is how dominating Linux is becoming in the middle tier and backend server space. Not only has Linux been killing Windows in this area but it is also killing mainframes, Unix, and is a favorite choice for grid computing.
Grid computing is an area where Linux makes the most sense to me. Companies like Google and Paypal are clustering thousands of cheap nodes or blades without having to pay a few hundred bucks per node or processor in operating system licensing fees. These companies are also taking advantage of the available source code and making tweaks to customize performance and security to meet their needs. Check out this article about how Paypal leveraged 4000 Linux nodes running RedHat and eliminated the need for an expensive mainframe. Here is a key quote from this article... In a mainframe environment, the cost to increase capacity a planned 15% or 20% "is enormous. It could be in the tens of millions to do a step increase. In [PayPal's] world, we add hundreds of servers in the course of a couple of nights and the cost is in the thousands, not millions.
I can personally speak to a real life business case for Linux. About eight years ago I worked on a project that had incredible data processing requirements. At that time, the only database technology that existed in the market place that even had the potential to meet our performance requirements was Teradata. They gave us a quote of $34M for the solution which was comprised of proprietary hardware and software. Back then, our entire IT budget was less then $34M. So we built our own solution which ran on a cluster of servers running Red Hat Linux for $100K. Throw in our labor and other fees and we spent close to $1M. What is more amazing is that we did not add a single employee to the staff to run the system since the system is self monitoring and self healing. With the Taradata solution we would have had to add DBA resources. This system is still running today and provides services for a product that generates over $100M a year.
I also stumbled across another article where IT shops are moving off of Unix to Linux for cost savings. A key take away from this article is this quote...Linux is the best-engineered, most interoperable platform for enterprise computing and is becoming the clear choice for organisations.
So we can debate all day whether Linux is the real deal on the desktop, but in the server world, Linux is king. The irony to me is eight years ago when I was proposing Linux servers (before Linux was cool!), I was getting the same push back and resentment that Linux on the desktop is receiving today. I am sure that three or four years from now I can dig out this post and joke about how people used to fight Linux on the desktop.
Back when Jaws was still considered a scary movie, the mainframe dominated the hardware marketplace. Well, Just when you thought it was safe to get back in the water, the mainframe or at least the mainframe mentality, is coming back.
Virtualization is as hot of a topic as BPM and SOA these days. Companies are saving millions of dollars by consolidating hundreds or even thousands of individual servers onto small clusters of servers serving up virtual machines. Other drivers for this technology are reductions in energy, emissions, and floor space, improved manageability, and easier disaster recovery strategies.
The Butler group published an article called, "The King is Dead - Long Live the Mainframe". If you have the time, this article is a great read. Here is a quote from the article:
We believe the wider adoption of the mainframe beyond these markets will be influenced by developments in the Service Oriented Architecture (SOA) paradigm, and the impact that the advancements in the capabilities of x86 server virtualisation is having in the market.One of the many reasons for the decline in mainframe usage over the years is the lack of products that are available for the mainframe platform. This is changing as Linux can now be the OS of choice on the mainframe. The article continues with this quote:
Another argument against mainframes has been the lack of commercially available software developed on the platform, which at best tends to be ported to the system at a later date, or not at all in some cases. This has created the ‘inhouse’ or customised solutions that have become associated with many mainframe implementations. However, since IBM announced support for Linux on its Z series this has become less of an issue.But even if companies are not considering mainframes as a platform for virtualizing their enterprise, one can't help but see the resemblance of today's virtual infrastructure with the mainframe infrastructure of the days gone by.
As I continue to research the virtualization movement, I keep stumbling across articles that point to various issues and challenges with virtualization. These range from security to inadequate monitoring and managing tools. When companies like VMWare and Open Source solutions like Xen resolve these issues, won't these solutions closely resemble the mainframe? If you think about it, the virtual server concept is basically the same thing as LPARs. The architecture behind the mainframes of yesterday are starting to look very similar to the architecture behind virtualization today.
IBM is using this opportunity to revitalize its mainframe sales. Most of their sales in recent years can be attributed to the fact that companies cannot afford the cost to migrate off of the years of legacy built on top of mainframe technology. Now, IBM can leverage the new mainframes running Linux as a solution to virtualization and Green IT initiatives. And by the way, the are eating their own dog food too.
