Partner Forum: IT's Crystal Ball - Does Your Monitoring Solution Predict Performance?

Andrew Chapman

By Andrew Chapman, VP of Product Management at CA Technologies

Earlier this year, we reached out to nearly 200 customers to learn more about how they position the mainframe in their datacenters today. Specifically, we were looking for trends around how they would like to improve service level management for their applications that span multiple platforms, including the mainframe. 

Looking at the results of the survey, it is very obvious that the mainframe is not viewed as a silo; it is not just a back-end, transaction-crunching or batch job-destroying behemoth in the corner of the data center. For many customers, it is a critical part of the infrastructure, supporting a growing number of systems.

We know that the mainframe still needs specialist care and attention so as not to compromise the promise of zero-downtime, the ability to meet onerous service level objectives and maintain the uncompromising security that we expect. Despite the mainframe’s need for specialized skills on a technical level, it still needs to be able to act like just any other component in operations management.

Our survey included IT management professionals from across all parts of the infrastructure; we asked respondents to consider how mainframe-centric their roles really were. The results indicate a trend towards roles that are now hybrid mainframe-distributed in nature.

Operations
Ninety percent of the operations managers surveyed managed some distributed systems, only 10% of them managed only mainframe systems.

Networks
For network management, 100 percent of respondents said that they managed network environments that bridged mainframe and distributed environments.

Line of Business Monitoring
Again, 100 percent of the line of business respondents indicated that their scope of monitoring was mainframe and distributed systems.

Applications
Seventy five percent of respondents said that their team managed applications that bridged both mainframe and distributed systems. More interestingly, none of the application programmers developed applications that ran only on the mainframe; they all have a distributed component to them.

The results above might not be a shock given that most mainframes are now deployed in direct support of some distributed environment. It is even less of a surprise when you consider that companies have matured away from focusing on the specific, underlying infrastructure and towards the need to support business application performance. We have not matured all of the way towards a focus on business application performance though. Consider that the majority of respondents still expect to use different tools to monitor the mainframe than distributed (65.1 percent). This makes creating a single business application view very hard to achieve.

Our feedback indicates that the number one ask from the community is the ability to have a single view of the performance of an end-to-end business service. Also, in the top three asks is the need to be able to perform IT triage as quickly as possible; identify the source of a failure to then allow the responsible parties to resolve issues in the shortest reasonable time. It is obvious that the industry is going to have to move towards solutions that offer true enterprise monitoring with predictive and forensic analysis to achieve these asks.

The ability to monitor a single system still has high value but the real benefit will come from being able to define a service, a set of systems or components within the IT infrastructure that delivers one or more business services and have the tools produce a consolidated view of their performance. Today, customers tend to focus on presenting that performance data in a way that facilitates a post-mortem for issues; this helps to establish what went wrong after the fact and then stop that from happening again. Our research shows however that the real value will come when we cannot only view the behavior of a business service in one view but also be able to predict when a component of the infrastructure will fail in the future. This converged monitoring and analytics will show that a system will fail if actions are not taken and then perhaps even propose the best corrective actions before that failure occurs. The tools and systems that support this vision are maturing, the community is seeing their value and workloads are only increasing, so it’s only a matter of time.

As the business line lead for the mainframe cloud and virtualization products group at CA Technologies, Andrew Chapman is tasked with growing the existing mainframe virtualization business and delivering new solutions based around leveraging the power of Linux on System z. He has been developing enterprise software since 2000 and is the author of a book on compliance as well as the author of a blog focused on the cloud, virtualization and the mainframe. Visit his website at www.MemorableURL.com.

Recent Stories
Security Corner: Confronting Mainframe Security

A Legacy of Wisdom: The Dr. John Ehrman Interview - Introduction

Message from SHARE: FUD Fighting