Run the numbers, and it makes economic sense to consolidate the cloud onto mainframes Get out your notepads and get ready for the mainframe resurgence. Tell your friends you heard it here. I’m not talking about a UNIVAC comeback, or Burroughs Large 2.0. I’m telling you, dear readers, that now’s a good time to invest in the few remaining big iron dealers: IBM, Unisys, Hitachi, and Fujitsu, baby. Why, you ask? Because mainframes solve the cloud’s impending challenges.High-performance? Check. Secure? Check. Worth their proverbial weight in x86 boxes? Check. The more businesses move into the cloud, the closer we’ll get to coming full circle. I’m talking Linux partitions on mainframes. Eventually the sheer number of x86 boxes involved will mean it makes monetary sense to consolidate. Some say cloud computing is the reincarnation of the mainframe. While I’m not sure that mainframe manufacturers would tell you they need reincarnating; I do think the cloud will soon be stickered “Powered by Big Iron.”[ Cut straight to the key news for technology development and IT management, with our once-a-day summary of the top tech news. Subscribe to the InfoWorld Daily newsletter. ] Back in the day, we started out with mainframes. Shocking for you youngsters, I know. Turns out mainframes were too large and expensive for basic computing applications, so the industry developed cheaper processors and put a server in everyone’s hands in the form of pizza boxes and x86 machines. Soon we learned that when any monkey can purchase and implement a new x86 server in a few minutes, datacenters get fat. To combat the sprawl, we patted ourselves on the back as we created virtualization. All we really created was virtual sprawl. We replaced the ability to easily add a physical server with the ability to point-and-click to spawn a virtual server. Funny how the 1960s mainframe could shoot some holes in both of these problems. Sure, they’re large and awkward, and they require a cooling system the size of my Honda, but they can do the work of an army of x86 boxes.I’m sure my hawk-eyed audience spotted my usage of the past tense above — especially since I italicized it. I started digging into the numbers and here’s what I found. You can purchase a Linux-capable IBM z8 series for as little as $40,000 plus an external disk array and network equipment. You can add additional IFLs for as little as $95,000, according to IBM’s 2008 pricing. Even at a conservative consolidation ration of 10:1, that already sounds reasonable. I know what you’re thinking: The z8 is two series old. So I followed up with some z9 series competitive analysis and found some practical, convincing customer testimonials. Allow me to summarize one: Jim Marshall (U.S. Air Force, retired) describes running two production applications on 45 virtual servers across three logical partitions. The applications were secured behind two firewalls and an authentication system. He priced this setup on both a z900 with one additional IFL and Intel x86 machines. The z900 with one IFL running Suse Linux, Novell e-Maintenance, and IBM 24/7 support cost $240,000. At $2,000 per server, the Intel machines, switches, firewalls, and zero support cost $840,000 — 71.5 percent savings sounds pretty good to me.Jim added that his Intel estimate would likely balloon to a million with support and software upgrades, while his z900 with 24/7 support “is not even breathing hard yet.” That last offhand comment is actually an excellent point. The z10 series is getting up to 10 times the performance on floating point and Java workloads versus the z9. The z10’s processors can run at 95 percent capacity, whereas x86 typically runs at 10 percent — maybe 30 percent if fully virtualized. That’s three times more capacity, even before we add virtualization to the z10.All that tells me that your datacenter doesn’t even need to be very large before you can start seriously thinking about a mainframe. I’m guessing your age will dictate whether you think I’ve just gone out on a limb or told you something you already know. If you cut your teeth on a mainframe, you know I’m right. If you’ve never touched a mainframe, you’ll think this is crazy talk. If you still think the Internet is the cloud, take a second to recall that the very first Web server running anywhere outside of CERN in Switzerland ran on an IBM mainframe on the Stanford University campus in California. Maybe we had it right the first time. The cloud migrating toward mainframes just makes sense. Related content analysis Azure AI Foundry tools for changes in AI applications Microsoft’s launch of Azure AI Foundry at Ignite 2024 signals a welcome shift from chatbots to agents and to using AI for business process automation. By Simon Bisson Nov 20, 2024 7 mins Microsoft Azure Generative AI Development Tools analysis Succeeding with observability in the cloud Cloud observability practices are complex—just like the cloud deployments they seek to understand. The insights observability offers make it a challenge worth tackling. By David Linthicum Nov 19, 2024 5 mins Cloud Management Cloud Computing news Akka distributed computing platform adds Java SDK Akka enables development of applications that are primarily event-driven, deployable on Akka’s serverless platform or on AWS, Azure, or GCP cloud instances. By Paul Krill Nov 18, 2024 2 mins Java Scala Serverless Computing analysis Strategies to navigate the pitfalls of cloud costs Cloud providers waste a lot of their customers’ cloud dollars, but enterprises can take action. By David Linthicum Nov 15, 2024 6 mins Cloud Architecture Cloud Management Cloud Computing Resources Videos