Bringing enterprise applications into production is a complex and difficult task. Web-based applications are designed with multiple architectural tiers (N-tiers), drawing on existing resources distributed throughout an organization and delivering them to users through a familiar Web browser. Multiple hardware and software platforms, security factors, and network infrastructures affect application availability, reliability and performance. Yet today’s enterprise applications must deliver 24x7 availability and unlimited scalability.
Before deploying core applications, organizations need to be sure that they will satisfy the demands of partners, suppliers, partners and customers. Testing, tuning and monitoring have become a critical aspect of any application deployment.
Good performance is a critical success factor of any application deployment, because poor response times can significantly impact user acceptance.
Generally speaking, the performance issues can be caused by -- REASONES
1. Network - Bandwidth and Latency
2. RDBMS and database server
3. Application itself
4. Hardware configuration of server and client
What ACTIONS could be taken before an application publishes?
1. Configuration on development databases with realistic data volume and profile, thus bad configurations are detected already in unit test phase.
2. Involvement of Subject Matter Expert for performance checkpoints in addition to the usual application quality reviews and for server and client sizing early before deployment.
3. Regular performance test cycles as part of any system testing, that cover new and old functionality, over client as well as all relevant interfaces.
4. How many locations will generate the load? How many users from each location?
5. How does the network performs? Should consider the VPN users.
PS: In general it was found that acceptable performance results can not be expected with latency higher than 100ms.
