IBM and the resurrection of the mainframe

Steve Mills runs IBM's $20 billion software business. He obsesses about large enterprises running thousands of transactions per second with terabytes of data and a need for absolute certainty of execution. It's a stack of enterprise software writ large, but don't get the idea that Mills has his head in the cloud. The 34-year IBM veteran doesn't put a lot of stock in the latest IT disruptor--cloud computing--for his customer base.

I met with Mills at IBM's Business Partner Leadership Conference in Los Angeles this week and asked the senior vice president of the IBM Software Group about his company's view on cloud (on demand) computing. "We have been running multitenancy [running multiple customers on a single machine with a single application instance] for decades and decades," he told me.

"It's a mainframe model where things run together but in isolation. The issue is whether the machines will bear up under the load of diverse work or will they grind down and you'll need to provision another machine. You need reliability, security, auditing, privacy, data integrity, automation and full isolation. You need to have a lot of layers in the environment."

In 2000, IBM resurrected the mainframe by bringing Linux and WebSphere to the platform and lowering the price of entry, according to William Zeitler, senior vice president of IBM's Systems & Technology Group. "You can build out a thousand smaller servers that need to be connected to ports and a fabric. You end up with a complexity crisis that has to be rationalized," Zeitler said.

As proof of its mainframe devotion, IBM is in the process of internally replacing 3,900 servers with 33 mainframes by the end of 2009, he added, noting that issues such as data center space and energy savings make the mainframe a cost-effective investment. Since 2000, units of installed mainframe capacity from all vendors has grown from 4 million to almost 13 million today, Zeitler said.

Mills dismissed the idea that Salesforce.com, NetSuite and others running enterprise applications are fully multitenant, meaning they run multiple customers with a single application and database instance with data isolation across a distributed network of servers and virtual machines. "This is a topic of much illusion. You can't change the laws of physics. Don't let anyone tell you that they have found the Holy Grail of multitenancy. Few will let you in to see if it is fully multitenant," he said.

Zach Nelson, CEO of NetSuite, responded in an e-mail to Mills' comments about multitenancy. "We don't run our data center on a single mainframe-like machine, we run it on hundreds of small machines," he said. "However, on each of those machines, we do support multiple customers on a single instance of the application and a single database instance (using virtual private database technology within Oracle). And the amazing thing is we are able to do this without re-inventing a single law of physics."

It could be that Mills and Nelson don't totally agree on the definition of multitenancy. Mills makes the case that software-as-a-service and multitenancy are and old concept that grew out of mainframe computing.

IBM launched "service bureaus" in early 1990s and continues to host about 1,000 customers today, he said. The hosting service ranges from running a whole company to partial outsourcing, as well as its services organization running a company's data center on premises.

"Everything goes back to the beginning, the mainframe. Today we have the same problem, but it's heterogeneous and distributed and not inside the box with a single control," Mills explained. He called MVS [mainframe operating system] a "40-year-old tour de force" that is packed with preventive code built up over decades that "fights like mad to keep machines running with extremely high reliability."

He contrasted IBM's industrial strength iron, whether on mainframes or racks of servers, with the infrastructure platforms from Google (App Engine) and Amazon.com (EC@). "Our corporate buyers are running mission critical apps, and they are not going to pick up their businesses and take them to some amorphous and ill-defined data centers," Mills said. "The cloud as some amorphous concept that meets all needs and requirements is science fiction."

That said, IBM is an arms supplier to the cloud vendors. IBM also has created a "Blue Cloud" initiative to provide data center infrastructure for Web sites. The rack-based iDataplex server is part of the initiative and is being tested by several sites, including Yahoo.

I asked Mills about entering the on-demand applications business (acquiring Salesforce.com or NetSuite comes to mind) beyond some parts of the Lotus collaboration suite. "With higher bandwidth and higher performance to make the remote seem local, the [market] opportunity has gone up. It's not clear that every company in the world sees it as an attractive next step to get IT function. It won't be one size fits all," he said.

At the Business Partner Conference, Ravi Marwaha, general manager of IBM Global Business Partners, said that emerging markets and the 1 million mid-size firms, with less than 1,000 employees, are key growth areas for the company. If IBM is serious about going after those sectors, which don't necessarily want to invest in hiring large IT staffs and building their own infrastructure, it will have to present a more clear strategy.

Mills said that IBM is happy to provide infrastructure, but not the enterprise applications. He explained that IBM gets its greatest return on investment via collaboration with the thousands of software developers in its ecosystem rather than competing with them. "There are enterprise apps such as SAP and thousands of others that serve the enterprise. For every $1 for their software licenses, there is $5 for related hardware and lots of services."

IBM is focused on working with its ecosystem to build more solutions--hardware+software+services. It is the trifecta that powers IBM's revenue and profit growth today. But what happens if the $5 becomes $4 and the $3 even less as customers move toward on-demand platforms that don't have all the complexity of today's IT hairballs that require IBM service interventions.

"We are constantly tracking where things are moving to with the IT automation providers. Our challenge is to make sure we sell to them," Mill said. "You take some Isaac Asimov a thousand years into the future where all businesses source services from service providers, but it's not likely to happen in my lifetime, my children's lifetimes or their children's lifetimes. There won't likely be only two companies left in business services. There is enough diversity for different business models to exist, and the service bureau model will get bigger and we can sell into it."