Analytics & Reporting

As remarkable as it is to replay a live production workload – with iReplay, that’s just the beginning. For every run, iReplay™ generates a comprehensive set of SQL-level performance measurements, using the same technology found in iWatch. These data, in turn, become the basis for detailed reporting & analytics.

Like iWatch, iReplay captures every SQL statement, and generates a long list of useful metrics: end-to-end response time; bytes in and out; network round trip time; server reaction time; total server time; and many more. Error messages are captured in their entirety, along with identifying details that may serve as a cross-reference for further drill-down, such as Process ID (SPID or PID). Even summary-level data is captured, exposing relevant trends and usage patterns.

Above, details from a trend-level report from the same set of runs, comparing Sybase 12.5
on a single-CPU machine to a 4-CPU machine. “Frequency” shows that the number of SQLs
executed is equivalent within 0.034% -- or 99.966% equivalent; run 2 (the single CPU machine),
showed average response times 60% greater than the reference run. The graph above
illustrates the performance pattern over the course of the same run.

To this rich data set, iReplay adds the vital dimension of comparison reporting. Users can generate detailed & customizable reports at the run level, session level, or transaction level – all with robust filtering. A unique algorithm in iReplay even matches queries from the reference run to the corresponding queries in the target run, allowing direct A/B comparisons.

This, in turn, allows iReplay to generate graphic reports that quickly answer the most vital questions concerning your tests.

Is the comparison valid? Was the number of SQLs executed statistically equivalent? Were the bytes in and bytes out similarly equivalent? Did all of the SQL constructs execute correctly? And how did performance compare? Did we get the 20% performance improvement our (database/OS/hardware) vendor was claiming? Are we really going to get the ROI we're expecting?

And the most critical question of all: are there any issues with the new (patch/release/hardware) that might prevent us from putting this into production?

In the end, it’s the ability to answer questions like these, with confidence, that makes i2Replay stand out – and will make it an indispensable part of your QA & testing process.