* Availability, security and data centralization are a few reasons why mainframe computing is alive and well
In a 2009 survey, 1,546 IT managers who are responsible for mainframe operations revealed their top management priorities. Read on for some best practices for addressing those needs.
A few weeks ago, I sat down with Bill Miller, president of the Mainframe Service Management division of BMC Software. Miller’s team had just concluded the analysis of data collected in its 2009 Mainframe Survey, which polled more than 1500 IT managers and executives worldwide who are responsible for mainframe operations. The primary focus of the survey was centered on the mainframe’s benefit to organizations in the tough global economy.
Despite the phenomenal rise of distributed computing over the past two decades, there’s lots of evidence that mainframe computing is alive and well and poised to support large businesses well into the future. There are still some applications that organizations just don’t trust to anything other than a mainframe; for example, global stock trading, or telecom operations.
Availability, security and data centralization continue to be among the top reasons why organizations prefer their mainframe to distributed systems. In fact, 62% of the survey respondents project their mainframe platform will grow and attract new workloads. Only a minute percentage of the respondents believe they should be looking for an exit strategy to take their enterprise applications off the mainframe.
IBM is doing its best to attract new users to the mainframe platform with its version of Linux for IBM System z (zLinux). This software allows you to run Linux and UNIX applications on the original virtual machine platform — the mainframe. You can run hundreds or thousands of Linux instances at once, all while simplifying your infrastructure. If BMC’s survey data is any indication, IBM can look forward to strong sales for its zLinux software, as “server virtualization” and “data center consolidation” are two of the top priorities for mainframe owners in the year ahead.
Related story: can mainframe use really grow?
Rounding out the top six IT priorities for the mainframe are: disaster recovery, MIPS optimization projects, application modernization and reduction of availability impact of planned/unplanned outages. Given his years of experience of mainframe service management, I asked BMC’s Miller for his best practices advice on what customers can be doing now to manage their mainframes to address some of these high priorities. Below are some of his suggestion pertaining to disaster recovery, MIPS optimization and capacity planning.
Disaster recovery: Being consistent in a time of change
Disaster recovery is always important, but its high ranking as a priority in BMC’s 2009 Mainframe Survey suggests that preventing data loss is even more critical in a tough economy. Perhaps the costs of data loss — in time, money, reputation and lost business — are even harder to bear when operating budgets are so tight.
Disaster recovery, of course, includes tactics such as having remote data centers, so businesses can switch over in the event of major disasters such as what occurred in the United States during the 9/11 terrorist attacks or during Hurricane Katrina, which happened in 2005. It’s critical to have a high degree of confidence that disaster-recovery programs will work.
In today’s heavily regulated climate, in which companies must document that they have taken all necessary steps to avoid risk, many businesses use capacity-planning models to test their disaster-recovery plans. These businesses model the transfer of all critical production work to the disaster-recovery systems and model how well it runs. If there is not enough capacity, the model shows the extent of the problems that will be created. The model can evaluate the risk, understand what the risk is, and plan enough capacity for handling a disaster recovery properly.
But equally important are robust practices for preventing small problems from becoming business disasters. For example, a common source of failure occurs when a bad SQL change is made by a database administrator. The outage may be localized, but it can still result in loss of business-critical data. Using modern data recovery tools, mainframe IT staffs can quickly isolate the source of the problem and initiate automated procedures for data recovery based on business rules. Such tools can speed mean-time-to-recovery (MTTR) and manipulate forward-recover points to reduce the amount of processing data that needs to be recreated.
MIPS optimization: Delivering big outcomes from small changes
For most businesses, the goal is to reduce MIPS consumption during peak times, so you can avoid having to overprovision (load up on MIPS) to accommodate peaks in transaction activity. Compounding the challenge is that predicting peak times is becoming more difficult, as the Internet and Web applications give customers the ability to transact business 24/7.
Inefficiencies in customer transactions could cause excessive MIPS consumption, a problem during off-peak hours but one that’s certainly magnified during peak hours. Ever-expanding peak windows make MIPS optimization one of the highest-value practices you can implement.
Using automated IT management technologies, mainframe IT teams can identify, prioritize and fix the sources of inefficiency in customer transactions. For example, a common source of inefficiency is application code that may have been written years ago in a different era of computing. The code wasn’t inefficient then, but it is now. And it doesn’t have to be a big piece of code to create big inefficiencies; it may be a small piece of code that runs many times.
Consider a large financial services firm that handles stock market trades over the Web for its customers. Financial news from Europe often creates a flurry of trading activity when the U.S. stock market opens, resulting in a deluge of trading transactions. This is good for the business, but is overly demanding on mainframe MIPS and the IT staff. Using modern IT management tools, the IT staff could identify MIPS-consumption sinkholes — for example, a small piece of SQL code that queries a back-end DB2 database as part of every customer transaction. By identifying and optimizing this small piece of code, the firm could likely substantially reduce its MIPS consumption.
By managing MIPS optimization aggressively and strategically, businesses can squeeze more work out of fewer MIPs, including accommodating new applications to support business growth. IT can delay the need for MIPS upgrades and their high costs in terms of software, energy, and people.
Mainframe capacity planning: Keeping ahead of the business
If MIPS optimization tells you how many MIPS you can save, good capacity management can project what those savings will mean for the business. Exactly when will you need to upgrade? And exactly what will you need to buy to serve the business effectively for the lowest possible cost?
Modern capacity management tools have predictive analytics, modeling and other sophisticated features that make it much easier to right-size the mainframe to business requirements. This makes capacity planning an invaluable tool in data center consolidation, server virtualization and application modernization projects. It can help ensure that the mainframe performance is optimized for any changes being made to the environment. An example would be if you are adding specialty workload-processing engines or moving more virtualized applications to the mainframe.
Capacity planning also can identify opportunities to save by moving applications from distributed servers to the mainframe, thereby ensuring that the business gets the best performance.
Like everyone else in business today, mainframe IT staffs are under tremendous pressure to save money. Automated management technologies exist for using the mainframe to save money while preparing the business for future growth. Leveraging automation technologies like the ones discussed above will help mainframe organizations meet current and future challenges.
