Technical database specialist.
I recently had some discussions about understanding a CPU load. My position on understanding CPU load is that the figures that are mostly used, which are CPU percentages and the load 1/5/15 figures, are not enough to get a complete understanding of l...
There is an ongoing debate about the usage of processes versus threads for programs that want to (or need to) parallelise work. In linux and most operating systems, processes are a completely separate entity, which means upon creation they get their ...
I have been writing bpftrace scripts to look at how tasks run on Linux in detail. One of these bpftrace scripts that I created takes the task states and shows the states for a given process. When I tried running the linux strace (system call trace) e...
I do love CLI. However, there are things that are better shown using graphical overview, such as a linegraph. A terrific example are the memory statistics in /proc/meminfo in linux. It's a long list, and even if you grep a few relevant statistics, yo...
This is a short blog introducing a docker build file that I created to quickly build a container for test driving PostgreSQL with wait event probes; see https://databaseperformance.hashnode.dev/timed-wait-events-in-postgresql about that. To create a ...
This is something that doesn't exist in PostgreSQL at the time of writing. However, it's extremely easy to add it, in a way that has negligible overhead. The overhead is the "classic" reason that PostgreSQL does not add it. The way to add timing is t...
Being able to understand where the query time is spent in PostgreSQL is a bit confusing. To understand what I mean by 'confusing', a fundamental concept that must be understood is the different statement execution phases of the PostgreSQL query engin...
This blog post is about an open-source tool that I created that can read node_exporter (Prometheus format) endpoints and print (performance) data just like the sar and other utilities would on unix-based systems. The main difference is the tool can r...
Lots of people investigate CPU performance using the way the traditional Unix tools such as sar, present these. That means that CPU usage for the different modes is shown as a percentage: 12:26:38 CPU %user %nice %system %iowait ...