A little more than five years ago (see “in the news“) IBM announced that it would begin installing and supporting the Linux operating system on its mainframe computers. A lot has happened with Linux and with mainframes since then, but first of all, what is a mainframe? Is it a “main” frame, a main “frame”, or something else? The wikipedia offers a history and perspective on mainframes, but my recollection is slightly different. IBM computers used to be constructed in steel “frames” that would fill a large building — and required a lot of plumbing to provide the circulating water to keep them from overheating. In one of the frames was a “console” — think of it like the keyboard of a PC — which provided many dials and switches — like an airliner cockpit — that enabled the “system operator” to control the computer and tell it what to do. That particular frame was the “main frame”.
Today’s most sophisticated mainframe, the Z9, stands a mere 6 feet 4 inches tall, weighs 3,836 lbs, and occupies a footprint of 27 square feet. I was in the room at the Hotel W last month when the Z9 was announced. It was quite impressive to see the sleek space-age system on stage with the power to replace thousands of separate servers. More than a billion dollars was invested in the engineering and development of the machine.
People have talked about the death of the mainframe for years but after seeing the Z9, you can be sure they are not going away for a very long time. In addition to the Z9, IBM announced an extension of it’s incredibly powerful virtualization engine software. The combination of the new mainframe and the new software will make it possible to turn a real datacenter into a virtual datacenter. This is a really big deal. CEO’s, CIO’s, and CFO’s are making plans to consolidate their datacenters using the new combo because virtual datacenters require fewer people, offer more reliability, and are much less costly to operate. Sounds good, but what is a datacenter and what is a virtual datacenter?
When you visit the web site of a company and click on something, a server in a datacenter somewhere gets the job of finding the web page or process you requested and delivering it to your browser over the Internet. A simple protocol called http (hyper text transfer protocol) describes exactly how the process works. With one user on the Internet and one server at the other end serving one web page it is quite trivial. With millions of users around the world visiting the web site at unpredictable times and making unpredictable requests for millions of documents, processes and transactions, it can become a nightmare for the people who are managing the datacenter.
The datacenter is the heart and soul of a modern day business. The datacenter is home for servers — computers which “serve” the web pages and processes that customers, business partners, vendors, employees, stockholders, and others are requesting. Ten years ago companies had a web server. Today corporate enterprises have thousands of servers — some have tens of thousands of servers. As new applications are added to the business, new servers get added to the datacenter. As demand grows, new servers get added to the datacenter. As companies grow geographically or through acquisitions, more servers get added to the datacenter. For historical reasons and to protect against disasters, companies and institutions of all kinds end up with multiple datacenters. At one point IBM had more than 100.
If the only challenge was managing large numbers of servers, things would be easier. That is not the case. Servers have magnetic disk drives for storage of programs, transactions and databases. The storage devices hold trillions of characters of information scattered among the servers. The servers must be connected to the Internet and for this purpose, datacenters have networking hubs, network switches, and routers. These networking systems are actually special purpose computers (as servers are) which connect the servers to each other and to the Internet. Many datacenters use their internal networks to implement network attached storage and storage area networks. Storage of data is increasingly done using storage servers which in turn have their own storage and software. And then there is backup using backup servers — with sophisticated tape drives that can store hundreds of billions of characters of information on a cartridge and computer controlled robots that keep track of which cartridge is which and when to do the backups. And then there are print servers to route the output of servers to various printers within the enterprise. This was an abbreviated version of how complicated things in real datacenters are.
To make things even more challenging, the enterprises of the world have a heterogeneous environment. That means they have PC’s, mainframes, new computers, old computers, big ones, little ones, IBM, Sun, HP, Compaq, Dell, Unisys, Windows, Unix, Linux, Cisco, Netgear, Nortel, and more.
How does all this get managed? Good question. It isn’t easy. An important part of the answer can be found at IBM’s Tivoli software division. Tivoli has created the “Performance and Availability Management” solution which provides a central nervous system that constantly gathers information about hardware, software and network devices, and, in many cases, cures problems before they actually occur. Using this kind of software is essential — but not adequate for the future. What is needed is a “virtual datacenter”.
Imagine looking through a window into a corporate datacenter (even though many of them are underground and have no windows) and you would see thousands of steel boxes mounted in six-foot-high racks with cables everywhere. You would also see people deploying “sneakerware” — people with track shoes on moving from rack to rack replacing things, changing things, and mostly adding new things. In an office somewhere is the executive responsible for making sure that everything is working properly and delivering the responsiveness to the customers that is expected. His or her to-do list is very long.
Now imagine a virtual datacenter. When you peer through the window you see three boxes — one is a mainframe server, another is labeled “storage”, and the third says “network”. There is a person at a large video console who is looking at what appears to be a dashboard. It shows a pictorial diagram of all the applications that are running in the datacenter on the mainframe — payroll, purchase orders, invoicing, web purchases, inventory management, training video streaming to new employees, etc. When one application area needs more server, storage, or network capacity the virtual datacenter automatically re-allocates capacity from another application area that currently has excess capacity. The virtual datacenter keeps resources balanced, and when a component fails, the virtual datacenter automatically allocates a spare or underutilized component to take over.
Sound like magic? Not really — it is a lot of software being created by teams among IBM’s thousands of programmers working in the company’s systems and technology and software groups. Collectively, the new software is called the virtualization engine. It is tightly integrated and optimized with IBM servers but also can include products from other vendors. It makes the real datacenter and all of its many thousands of components appear virtual — and simple. By turning the real into the virtual and providing tools to manage the resulting virtual datacenter, the virtualization engine will allow management to get their arms around what has been a very challenging task. The result will be that businesses (and governments and hospitals and universities) which use virtual datacenters will be able to be businesses on demand in the real world.
It is the evolution of the mainframe that is making virtualization practical on a very large scale. If you are interested in reading or participating more in the discussion about mainframes, take a look at the new mainframe blog.
