Welcome! | Sümer Cip

May thy bits chip and shatter: Patterns for Building High-Performance Observability Pipelines at Scale

Mon, 30 Dec 2024 18:48:06 +0300

Motivation

As seen from the title, I’m a huge Dune fan. Much like sandworms lurking beneath the surface, ready to appear unexpectedly, the scale and complexity of observability data can be just as massive, unpredictable, and a little terrifying.

“But I do see a way. There is a narrow way through.”
– Paul Muad’Dib Atreides, Dune Part Two

Sci-fi analogies aside, designing high-throughput observability data pipelines is a fascinating intersection of datastore design and architecture, packed with unique challenges.

In the Making of Python Fitter and Faster

Fri, 04 Oct 2024 18:49:37 +0300

Python’s Performance Revolution

Since Python 3.11, there has been a strong, ongoing effort to make Python faster, and the results are clear. Performance improvements are real, and the work continues.

It’s both refreshing and surprising to see such significant speed gains in a language that’s nearly 30 years old.

But make no mistake: there’s no magic here. In my humble opinion, the success can be attributed to a few key factors:

Articles

Sun, 31 Dec 2023 17:47:24 +0300

Here are some articles and blog posts mostly related to my work found at other sites:

2023

2022

Challenges of Asynchronous Python profiling

2020

Talks

Sun, 31 Dec 2023 17:47:24 +0300

Here are my recorded talks from various conferences mosly related to performance and observability:

2025

FOSDEM 2025 - Go-ing Easy on Memory: Writing GC-friendly Code

Slides: https://docs.google.com/presentation/d/1qP9RuO0BOGC6spwwdeX9lFdZW_jv6XI0ZIhjFh3dVLU

2024

GoKonf Istanbul 2024 - Go-ing Easy on Memory: Writing GC-friendly Code

2021

2020

PyCon Turkey 2020 - Hunting Performance in Python Code - (Video not available, but details can be seen here)

Inside the Go CPU profiler

Mon, 26 Sep 2022 18:42:25 +0300

Go is one of its kind when it comes to profiling. It includes powerful and opinionated(!) profilers inside the runtime. Other languages like Ruby, Python, or Node.js, contains profilers or a few APIs for writing profilers. But they are limited in scope compared to what Go offers out of the box. If you want to learn more about the types of observability tools that Go provides, I highly recommend Felix Geisendörfer’s The Busy Developer’s Guide to Go Profiling, Tracing and Observability

About

Fri, 31 Dec 2021 17:47:24 +0300

Hey 😃,

My name is Sümer Cip. I have been coding roughly my whole life and have worked for various companies in different Software Engineering roles. Visit my LinkedIn profile for more information. I have tried my best to contribute to various Open Source projects whenever possible, most notably Yappi. I have always been passionate about developing tools to simplify/optimize systems. These days, I am mainly interested in Observability, Distributed Systems, Database internals, Go and Python.

Improve zsh startup time via lazyload

Fri, 31 Dec 2021 11:29:57 +0300

My shell is slooow!

This is my first blog post for years. I have been finding excuses for not writing one over the years, but I feel this is the right time.

I was coding as usual when I started something odd with my shell. I have done few adjustments to my .zshrc file and out of a sudden it became very slow to spawn a new shell. As a person who has a obsession with performance, it became my duty to find out the problem.