Increasing file I/O performance for JVM on Linux

Day 2 /  / Track 3  /  RU / Hardcore. Really hard and demanding talk, you'll understand only if you're an experienced engineer.

Analytical platforms process more data and the question about their performance arises more often. Each platform has its own optimization methods and techniques but to create a truly high performance system we must have a deep understanding of all constituent elements of the platform and data flows. The most expensive and frequent operation in the data delivery system is working with the file system and it needs to be optimized.

In Dmitry's talk we will consider file I/O optimization methods and data copying and how this can increase data transfer throughput by 20%.

Download presentation


Dmitry Bundin
Dmitry Bundin
Grid Dynamics

Dmitry has been in IT since 2014. Over the past few years he has been actively developing I/O intensive applications on Java, C/Linux and related performance issues. Prior to that he developed banking software and a data processing platform on Spark. He has experience on working with Scala, functional programming and typelevel stack. Currently works as Big Data developer at Grid Dynamics.

Invited Experts

Sergey Melnikov
Sergey Melnikov

Sergey works on a high-frequency trading application at Raiffeisenbank since 2016. Basically, he's into speeding up the existing and writing the new low-latency Java code. Before that he worked at Intel as an engineer for the compilers' performance for C / C ++ / FORTRAN languages.