Scalability for Web applications
What is scalability?
In simple terms, the server should be able to meet the increased demand for HITS per SECOND/ Transactions per Second over a period of time in increasing user ratio, without failure.
If the Hits per second decreases at some point of time, then it is indicative of decreased performance. Typically a decrease in scalability can be traced to database indexing problem, a full data cache or a saturated network connection.
Therefore it becomes necessary to test the disk usage, memory, or network components while stress testing.
What is perfect scalability?
With increased number of users (X axis), the Hits per Second parameter also keeps on rising proportionally. This indicates perfect tuning between various components of the application.
This trend will continue until we reach a point where further increase in user load will see decreased activity in terms of “hits per second” and “transactions”.