Application Response Time SLA´s: Lessons Learned by a Major Outsourcer

This presentation provides an overview of the lessons learned by a major US-based outsourcer when faced with the challenge of measuring and reporting on end user response times, to satisfy contractual service level agreement (SLA) commitments, for the mission-critical applications of its largest client, a US automobile manufacturer. The presentation explains the reasons for adopting what was then a new technological approach to meet this challenge and the response time metrics that the outsourcer and its client determined would be used to measure service level compliance. This paper first discusses the different methods for measuring end user response times in enterprise TCP/IP networks. It compares a relatively new passive monitoring technology with traditional methods that the outsourcer evaluated and provides the advantages and disadvantages of each. The second lesson learned dispels a common perception about the distribution of weighted average response times, and proves that theoretical projections are no substitute for observing real transaction response times. The third lesson explains how the outsourcer and its client arrived at a single metric they would use as the basis for measuring end user service level compliance. Finally, the paper briefly discusses the political and cultural issues that must be overcome when implementing this kind of service level measurement approach, and tips for getting "buy-in" from IT groups who are accustomed to using traditional approaches