Documentation

Performance Manaement with DBpulse

Download Full White Paper.

INTRODUCTION

Application performance management is a complex discipline. There are so many layers of software, hardware, and networking that it is hard to pinpoint where problems lay. One area that is consistently problematic is DBMS performance. The main reason for this is because the DBMS is a bottleneck. PCs and application servers can be horizontally scaled, however, database servers cannot. In addition, the most common interaction with DBMS servers is through SQL -- a high-level, non-procedural, interpreted language. This makes the DBMS appear like a black box where queries go in and result sets come out. How it actually works is a mystery to most, and tuning a DBMS is even more mysterious.

Problem identification is the most difficult element of a tuning engagement. In many situations, users complain that the application is slow, but they cannot pinpoint exactly where it is slow. A tuning specialist spends the vast majority of her time interviewing users and watching the system run. With luck, she might be present when a performance problem occurs and can try to track down the queries that caused the slowness. Otherwise, she spends countless hours waiting for a problem that never happens or is hard to reproduce.

So, how can one go about systematically improving DBMS performance and alleviating poor response times? The simple axiom is: in order to improve something, you have to be able to measure it. But, how can DBMS performance be measured? What are the proper metrics and how frequently do they need to be sampled? And how will the measurement affect overall performance? This document addresses these specific issues and introduces a revolutionary new service from Foresight Automation that will detect and correct most database performance issues and keep them from re-occurring.