Operating Systems

A definition for IT strategy recently found on CIO Index (www.cioindex.com) says that it’s an iterative process to align IT capability with business requirements. Key is the alignment of business and IT capability. This assumes that business drives IT and not vice versa.

Over the years, we’ve had many data center managers, CIOs, operations managers, and other IT management folks ask the question, “I can run Linux almost everywhere, but how do I decide when to run it on the mainframe?” At that point, most of them say something like, “Our architects don’t understand the strengths of the mainframe, so they will never define an application that runs on Linux on System z or even on an IBM zEnterprise BladeCenter Extension (zBX).”

You should run applications where it makes most business sense. For manageability, security, reliability, stability and recoverability, you run on a mainframe. And if it happens to be a distributed-type application, you may be able to run it on Linux on System z. But how do you decide if something makes “business sense”? Maybe it’s better to look at it from another angle: What doesn’t make business sense? Let’s look at a few examples:

• Mediocre security. Nobody who sells something wants to ship goods that aren’t paid for, and nobody wants to read in the newspaper that a hacker has stolen the personal data of 10 million people from your server.
• Manageable infrastructure. Business people know that a simple solution is often the best. Show them a large data center with lots of devices and cabling and they will love to show it to their clients. However, in their hearts, they would rather see a large black box with one on/off switch. It gives them confidence.
• Lack of reliability and stability. Business people are in the business of selling things. Prevent them from selling by taking away the tools they need to sell and they will act hostile. They expect the business tools IT provides them to be 100 percent reliable and always available.
• Unrecoverable applications. Try to explain to a business person who just publicly admitted their client-facing application has been down four hours, that “we can’t guarantee 100 percent all transactions have been recovered.” This probably won’t set well.

A Match Made in Heaven?

When IBM announced Linux on System z, they wanted to offer the combination of the flexibility of Linux and the reliability of the mainframe. This was a match made in heaven, especially for those IT managers who understand that only an IT strategy without technical bias and driven by the business is sensible. So, why does it take so long for some to understand that Linux on System z really is one of the best possibilities to fulfill the requirements of the business? There are a number of reasons for this:

• Many people think that running applications on Linux on a mainframe is like going back in time. “We tried everything to get off the mainframe, and now we’re trying to get stuff on it again?” These people clearly have no understanding of the concepts of today’s modern mainframes, the scalability they offer, and the Total Cost of Ownership (TCO). They turn a business discussion into a technical discussion, knowing they can confuse business people enough with technical jargon so they can do what they think is best.
• Others have the opinion that Linux is open source, so it can never run strategic business applications. They don’t consider that the Internet would stop working if all open source software was removed today.
• Some will say that it’s so difficult to make the TCO calculations because “we don’t know how many Integrated Facility for Linux (IFL) processors we may need.” This has never stopped people from upgrading a Customer Relationship Management (CRM) or Enterprise Resource Planning (ERP) system, even when they don’t know what the hardware requirements for the next release will be.
• And there’s always the category that simply says “our architects have decided that hardware platforms X and Y are the corporate standard and the mainframe is just ‘tolerated.’ ”

The New Normal

Let’s get one thing clear; the world has changed dramatically in the past four years. What was “normal” four years ago is now often considered extravagant, and the “unthinkable” of four years ago is now often considered “the new normal.” IT managers who are close to the business feel this every day, and those in IT who’ve been shielded from all this news will try to continue what they did. Our job in IT is simple: Get the business the software it needs, run it in the most efficient way, and don’t limit the growth of the business in any way. Every unnecessary dollar/euro spent on IT can’t be spent on anything else.

In this light, moving software that currently runs on more than 50 underutilized distributed servers to Linux on System z where utilization rates of 95 percent are considered normal makes good business sense. So does moving applications that already have a strong connection with data hosted on z/OS. By bringing these applications onto the mainframe, you can reduce network components, increase performance, and improve security.

Linux on System z in Action

Take, for instance, a large hotel chain that operates around the world. They had a simple wish list: improve services, reduce license cost for databases and the application framework, and leverage the existing footprint of their automation and security software. The solution they implemented helped them better control costs and implement Linux on System z to deliver the reliability they’ve experienced on the mainframe. They are confident they will achieve savings of 1 million euros ($1.27 million USD) in the next 18 months.   

Let’s not forget that once your software runs on a system that’s well-known for its reliability, availability and serviceability, it needs to be managed with software that supports those same three points. In other words, the two main reasons to move applications to Linux on a mainframe are cost-savings and improved manageability. The latter can only be achieved using the same type of software that has managed mainframes for decades, but with a variant that supports Linux. Better still, if it’s integrated with the software you already have to manage the mainframe, the savings on management costs will be even greater.

There’s no secret formula that would tell you exactly what part of your stack can be moved to Linux on System z to achieve maximum savings with a minimum amount of investment. However, talk with your peers who have already made this move. Better still, take a step back from the day-to-day technical approach of the whole problem and change it into a business discussion.

This discussion won’t be easy, as emotions will run high and your technical people will have to let go of their bias. It may even require you to say that you may have been wrong (or “not completely right”) in the past. However, the facts about mainframe costs have significantly changed in the last decade. Seek advice from a trusted, platform-neutral partner; they can help you uncover important factors that influence business decisions and tell you about the expected results when running it on another platform. Find a partner that doesn’t have any technical bias and is willing to act as a referee and a diplomat more than a technical advisor. The technical decision isn’t hard; the “softer side” is much harder. If you’re serious about moving software to a platform that offers your business better RAS and you want to save some money while you are doing so, seek out that partner’s advice.