APM Data Center - Performance Best Practices that can change your business

W H I T E PA P E R
BEST PRACTICES: APPLICATION PERFORMANCE management

DATACENTERW H I T E PA P E R : A P P L I C AT I O N P E R F O R M A N C E M A N A G E M E N T
Datacenter best practices
From its roots as a mainframe performance discipline as DB2, CICS, Websphere MQ, VSAM, etc. Very quickly we arrive at a collection of "element" or platform tools needed to monitor a complex, implemented throughout the Systems Development Life if monolithic, system.Cycle (SDLC), Application Performance Management Generally, these tools do a good job at high-level resource monitoring and (APM) has become increasingly important as end users offer more granular and detailed monitoring of specific subsystems. For rely on ever more complex applications to enable critical the DB2 or CICS practitioner, for example, they provide good parameter-level tuning feedback, identifying environment-specific performance business transactions. Having expanded beyond the enhancement opportunities. The collection of relatively mature tools mainframe environment into the distributed, web-enabled on the mainframe has led to a well-defined discipline for resource world, APM has gained a necessary end-to-end context, monitoring. Organizations that have implemented proactive performance management programs-sometimes referred to as MIPS Management focusing on the identification, prioritization and resolution (MM)-have been served very well, realizing significant cost savings by of performance and availability problems affecting avoiding CPU upgrades to support inefficient applications. While the business applications. primary driver has been cost reduction and cost avoidance through MIPS reduction, the adjunct benefits often include faster-executing applications In our introductory paper, we introduced the concept that APM can be and improved code quality.considered fundamentally as a collection of infrastructure monitoring tools. Because monitoring relevance is achieved via an understanding While a primary concern remains the need to reduce MIPS usage of the path the transactions take through the system, it is valuable to wherever possible, there is a growing emphasis on end-user experience focus our monitoring on the health of the infrastructure components that (EUE), driven by the technology shift toward more complex environments support the application. Likewise, business relevance is achieved via an (e.g., web, SOA, EDI) and the organizational need to better align IT with understanding of end-user experience, so we need to understand how business goals. No longer is transaction response time directly related to well the application serves the users. And, finally, because diagnostic mainframe subsystem performance; an application using DB2 cannot be relevance is achieved by the correlation of end-user experience with assumed to be performing well just because DB2 resources are available. infrastructure monitoring along the transaction path, we certainly need to And today's business goals emphasize performance-as experienced correlate poor end-user experience with the infrastructure root cause. by the end user-on a par with cost containment. Here is where most mainframe performance management solutions fall short, as they focus In this paper, we'll examine-in datacenter operations that involve the primarily on resource monitoring, resource tuning and fault correction. integration of mainframe, distributed and web-based systems-how APM This limitation parallels what we see in the distributed world-a collection influences, and is influenced by, the generally more structured tools and of tools often fails to identify application performance problems as processes available on the mainframe. perceived by the end user.With an eye toward APM, we can extend the value of these mainframe Resource performance management tools by adding or emphasizing EUE as a driver for tuning. This to Virtually every company running z/OS (mainframe) systems practices some degree flips the equation. By managing and tuning application some level of performance management, choosing from a wide variety performance as measured by end-user response time, we may of tools focusing on system and resource monitoring. Generally, these decrease the overall resource requirements of the tuned applications tools leverage SMF data to provide monitoring, alerting and automation and transactions. However, the primary goal of... [download for more]