7–9 November 2016: Conference & Tutorials
9–10 November 2016: Training
Amsterdam, The Netherlands

Death by a thousand paper cuts

16:10–16:50 Tuesday, 8/11/2016
Metrics/monitoring Middleware, Networking G102/103 Audience level: Intermediate
Average rating: *....
(1.78, 23 ratings)

What you'll learn

  • Gain awareness of one more source of performance problems: a multitude of fast operations

Description

When you have a performance problem to solve and your monitoring/profiling shows that out of 10 seconds of user request processing, you spend 8 seconds running a single database query or talking to an external REST system, then the problem’s root cause is clear, and the solution is simple: optimize that one slow query. But in many cases, there is no such obvious root cause.

Nikita Salnikov-Tarnovski presents several examples in which SLAs failed due to a multitude of fast operations, such as database queries, garbage collection, and file attribute checks. Nikita describes the challenges of applying the usual monitoring and profiling techniques when trying to detect such problems and explores how such problems can be spotted, verified, and remedied.

Photo of Nikita Salnikov-Tarnovski

Nikita Salnikov-Tarnovski

Plumbr

Nikita Salnikov-Tarnovski is the cofounder of JVM performance-monitoring product Plumbr, where he contributes as core developer and team lead. Besides his daily technical tasks of solving various problems, he is an active blogger, a three-time JavaOne Rock Star, and a frequent speaker at conferences such as JavaOne, Devoxx, GeeCON, JFokus, GeekOut, JPoint, and JEEConf.