The open-source operating system gets a Powerful boost.

In my last article, "Starting Your Sojourn into Virtualization on the Linux OS," I discussed the "no additional purchase necessary" virtualization that is embodied in the Linux Kernel-based Virtual Machine, referred to as KVM. For single-server use, such as I described with my co-located Web server, KVM is all that's required to take full advantage of the powerful commodity server hardware currently available. Larger shops, which need the bells and whistles of more sophisticated virtualization (physical machine pooling, guaranteed performance per server, etc.), generally move up to VMWare and its enterprise products, with their matching enterprise price tags.

Either KVM or VMWare are capable of running any main-stream Intel-based operating system. Should the desired guest operating system be Linux, shops running a sufficiently large Power machine have the option of carving out a partition and spinning up an instance of Linux in fairly short order. The only caveat to this option is that the Linux distribution has to be compiled for the Power platform. Both Red Hat and SuSE meet that requirement, and both are fully enterprise-quality distributions. Other distributions, such as the Red Hat spin-off CentOS, are hit-and-miss with regards to the Power platform, so those who favor the alternatives should check for availability.

Up to this point, I would have directed clients toward IBM's xServer line for new hardware to run Linux workloads. The products are well-engineered, readily available from an IBM Business partner or online mega-warehouse, and fully supported by both IBM and the aforementioned enterprise Linux vendors. And I never had heartburn over any client wishing to go the Power route either, provided that they already had a Power system in-house, even if that meant buying another Power system (or upgrading the current one) to accommodate the new load. The thought of recommending to a client that they buy into the IBM Power line strictly for Linux workloads never crossed my mind. That is, until now. On April 25 of this year, IBM announced the new PowerLinux solutions. The game has changed significantly.

I'm making an assumption that most of the readers of this column and Web site work in shops where IBM midrange computers (S/3x, AS/400, iSeries, etc.) have a presence. If your midrange career has been a long one, as has mine, then you've seen the developments that IBM has put into the platform. While their marketing efforts may have been somewhat erratic, their engineering efforts haven't been. Successive generations of the hardware have gotten progressively smaller in size, while at the same time increasingly powerful. On the financial side, the cost for purchasing and maintaining these systems has fallen along with the physical size. It never fails to amaze me how much bang for the buck these IBM systems offer.

Of far more interest to me than financial considerations is the reliability and manageability of the systems under my care. Here the midrange boxes shine. My first encounter with an IBM box was fresh out of college; it was an IBM System/34. The thing never died. In fact, the first time I saw the lights flicker while I was in the machine room, I cringed, as I expected to see the system crash. It never did. The power supply was over-engineered to the point that the power supply capacitors were able to carry the system for that brief interval. And we've all heard the story about the "lost" AS/400. You know the one that was supposedly hidden by renovations for a few years until a service tech came to look at it because no calls had been placed on it in ages? I'm not sure if the story is true or not, but if you're like me and have a history with these machines, it's not inconceivable. So the reliability isn't in question. Neither is manageability, as IBM has invested a lot of effort in developing its management software suites, allowing smaller IT departments to handle more-demanding human workloads.

Today, with the Power 7, IBM has converged product lines into a single hardware platform, so their manufacturing, engineering, and support costs have dropped due to fewer product lines. The evidence is clear as the price/performance ratio has been improving, as I alluded to earlier, so we all benefit. The weak spot in IBM's portfolio, though, has been for Linux-based workloads. Given that Linux runs on Intel boxes, and indeed Intel servers are where the lion's share of Linux instances live, IBM had to compete with the likes of HP and Dell for rack space in data centers. And as any of us involved with purchasing know, the bean counters are always looking at the bottom line. So why buy an IBM server when Dell is offering one with equivalent horsepower at a lower price? While I'm sure that the managers at IBM aren't thrilled about losing a server sale, even if it's just a low-powered, commodity, off-the-shelf configuration, I'm equally sure that they are even less enthused that their competitors are getting their feet in the door. Those bean counters may start listening to the siren song of Microsoft and wonder why they're paying a premium for those other machines from IBM, running an OS that isn't what they're always hearing about. I've seen this happen many times, and judging from the comments on some of the midrange mailing lists, I'm not the only one.

The genius of the new PowerLinux systems is that IBM has managed to create a system that allows existing Power customers a means of adding more capacity to existing data centers at a lower price point, allowing them to offload their Linux workloads from their existing Power boxes onto a system "tuned" for that purpose, thus freeing up the more expensive Power 7 resources. Those resources can then be put back into a pool for additional i5/OS or AIX instances or for adding additional horsepower to existing ones. Better still, with the new PowerLinux systems in the portfolio, IBM can attract new customers wishing to add Linux workloads, at a price/performance ratio that competes extremely well against their Intel brethren. If the IBM press releases are to be believed, the PowerLinux servers will outperform Intel servers by a significant margin. Comparing a machine that uses a more powerful processor designed for a specific operating system versus a machine that has a less powerful CPU and is designed to be more "generic" would intuitively lead one to believe that the machine with the better processor, designed for a specific workload, is going to win, hands down. I'll tend to believe the press releases. Besides, it's already been demonstrated by Microsoft, with its XBox Gaming platform, and by Apple, whose hardware has been "optimized" for use by Apple's operating systems. The icing on the cake is that new or existing customers can plug into all of the nice management tools that IBM has successfully demonstrated in its current Power 7 line. And PowerVM hypervisor precludes the need for those pricey VMWare enterprise virtualization solutions.

It looks like IBM has finally turned its attention to an area of the marketplace that it has not been able to compete in as well as it should. Truly, I predict that IBM has hit a home run with the PowerLinux platform and that it will keep the barbarians (read: HP and Dell) at the gate. Bravo!

Barry L. Kline is a consultant and has been developing software on various DEC and IBM midrange platforms since the early 1980s. Barry discovered Linux back in the days when it was necessary to download diskette images and source code from the Internet. Since then, he has installed Linux on hundreds of machines, where it functions as servers and workstations in iSeries and Windows networks. He co-authored the book Understanding Web Hosting on Linux with Don Denoncourt. Barry can be reached at This email address is being protected from spambots. You need JavaScript enabled to view it.