How we improved our Flame Graph rendering by moving computation to the backend.

Parallel Flame Graph Rendering

And why I don't want my database to choose the best encoding for me (yet)

Das Problem mit German Strings

Using eBPF to record your programs dying breathes

OOMProf - Profiling on the Brink

Collect and upload profiles -- even without a network connection

Offline Mode for the Parca Agent

How Off-CPU profiling works and how to get the most out of it

Introducing Off-CPU Profiling

A step-by-step guide to achieving reproducible builds with Next.js.

Reproducible Builds with Next.js: A Practical Guide

Introducing Remote MCP Support in Polar Signals Cloud

The MCP for performance engineering

How we designed our database for complete control over concurrency, time, randomness, and failure injection.

Deterministic Simulation Testing in Rust: A Theater Of State Machines

Application Performance engineering debt could live a long time without noticed and cause a significant overhead in developing new features, performance and the cost of operation.

How Continuous Profiling reduces Engineering Debt and increases Application Performance

Bringing what we've learned to our next generation database.

New year new database

Best Laid Plans

Attending ai.engineer 2025 in San Francisco

Latest in Continuous Profiling for GPUs

A deep dive into the time-aware visualization that complements traditional flame graphs in performance profiling.

Flame Charts: The Time-Aware Sibling of Flame Graphs

Exploring how well the various forms of PGO work

Sampling PGO

How we added a Markdown editor to our Sanity setup and other improvements

Enhancing Our Writing Workflow: Adding Markdown Support to Sanity CMS

Why profiling data is low-risk, high-value observability

Is Observability Data Sensitive?

Record and view profiling data from GitHub Actions runs easier than ever before!

Introducing Profiling GitHub Actions

Correlate profiles with arbitrary application-specific metadata in async Rust applications

Correlate Profiling and Distributed Tracing in Async Rust

Making Polar Signals Pricing more predictable and transparent

Updated Pricing for Polar Signals Cloud

Introducing Continuous GPU Profiling: Get Performance Insights Over Time

Maximize GPU Efficiency: Visualize, Analyze, and Optimize with Precision

Case Study: How turbopuffer leverages Polar Signals for Continuous Profiling

turbopuffer: Consistently High Performance at Scale

"High tail latency for one of our customers, was diagnosed immediately by seeing a large On-CPU time span" 

Unlocking Efficient Memory Profiling and Faster Debugging

“68.37% of CPU [...] with a one-line code change [...] went down to 31.82%” 

Doubling Throughput with Continuous Profiling and Optimization

Reinforcing our unwavering commitment to robust security

Polar Signals Achieves SOC 2 Type 2 Compliance

kubezonnet: Monitor Cross-Zone Network Traffic in Kubernetes

How we scrape callstack information from the LuaJIT engine for profiling

A Walk with LuaJIT

Filter and aggregate profiles by whatever dimensions make sense for your application

Custom Profiling Labels for Rust, Go, and C++

Polar Signals has achieved SOC2 Type 1 compliance, reinforcing our commitment to data security and customer trust.

A Milestone Achieved: Polar Signals is SOC2 Type 1 Compliant!

Polar Signals Is Organizing PromCon EU 2024 in Berlin!

Polar Signals Organizing and Attending PromCon 2024

A new way to streamline profiling with predefined label names

Improved Views on Polar Signals Cloud

A new and easier way to filter profiles on Polar Signals Cloud

Introducing Simple Querying Mode

Provides an overview of the processing status of your project's debug artifacts, such as debuginfo, executables, or sources.

Debug Artifacts Status Page

The Parca project is embracing OpenTelemetry

Parca + OpenTelemetry = Bright Future

Comparing Profiles is even more helpful now!

Allows you to aggregate your profiling metrics across processes by grouping them by specific labels.

Introducing Metrics Aggregation

A new intuitive way to explore the control flow through specific functions in your application.

Caller / Callee View in Table panel

We've revamped our filtering by binary feature to be more intuitive and functional!

New and Improved Binary Filtering in Icicle Graphs

How we used WASM and some Go runtime modifications to run deterministic simulation tests against FrostDB

(Mostly) Deterministic Simulation Testing in Go

Attending KubeCon EU 2024 in Paris, France.

Polar Signals at KubeCon EU 2024

How we connected the worlds of tracing and profiling using Parca Agent, the eBPF-based profiler.

Correlating Tracing with Profiling using eBPF

Insights into how much profiling data your projects are sending for an even more predictable billing.

Introducing Billing Usage Metrics

A Collaborative Leap Forward with Canonical

Embracing performance profiling: Canonical enables frame pointers in Ubuntu 24.04 LTS

Why dynamic dispatch is necessary and how to fix it with devirtualizing and PGO.

The Cost of Go's Interfaces and How to Fix It

Scope requests with pre-defined filters and gain precise control over your profiling data!

Announcing the new "Views" feature on Polar Signals Cloud 

Preview shared profiling data and keep a history of your shared profiles!

Announcing New Features on pprof.me: Previews and User History

Profiling Arm64 with eBPF in Parca Agent

We're ready to share our product with the world!

Polar Signals Cloud is Generally Available

Dive into the internals of profiling interpreted Python and Ruby code using eBPF

Profiling Python and Ruby using eBPF

A look at profiling Modular's Mojo language.

Profiling Mojo 🔥

A look at an exciting new feature in Polar Signals Cloud: Source View.

View Profiling Data within Code

Or: A tale of handling unexpected behaviour by the v8 runtime.

Correctly Profiling Node.js with Zero-Instrumentation

Attending KubeCon EU 2023 in Amsterdam, Netherlands.

Polar Signals at KubeCon EU 2023

Say goodbye to taking screenshots of flame graphs/profile data. Now you can embed the interactive profiles directly in any website.

Embeddable Profiles in pprof.me

Flame graphs and icicle graphs are a great way to visualize performance profiles. In this post, we will learn how to read and interpret them.

Ice and Fire: How to read icicle and flame graphs

How we improved our database query performance by 80% and the intuition behind the changes that led to this improvement

Designing Your Indexes for Better Database Performance

We have exciting updates in this release, including improved JIT symbolization, reliability and quality improvements, step querying, and more!

What’s new in Parca 0.16 and Parca Agent 0.13?

Symbolizing Postgres Profiling Data

Polar Signals at FOSDEM 2023

Parca Agent is our open source, always-on eBPF-based continuous profiler. In this article, we will learn some of the concepts behind Parca Agent and use it to profile our system.

Introduction to Parca - Part 2

Looking at some exciting features that recently landed in Parca.

What's new in Parca?

The tradeoffs and considerations building continuous, fleet-wide profilers.

Design Decisions of a Continuous Profiler

Deep dive into the new DWARF-based stack unwinder using eBPF.

DWARF-based Stack Walking Using eBPF

Open-Source Parca Agent is adding support for native applications compiled without frame pointers.

Announcing Profiling Support Without Frame Pointers

Parca in Grafana without having to install a plugin!

Parca now also in Grafana Core

Attending KubeCon NA 2022 in Detroit, USA.

Polar Signals at KubeCon NA 2022

Parca Plugin for Grafana!

Good news! pprof.me, our free profile sharing service, has got a shiny new look and feel. It's now easier than ever to share and visualize your profiles with the world.

Improved pprof.me! 🎉

Making code faster without modifying it using Profile-Guided Optimizations.

Exploring Go's Profile-Guided Optimizations

An example of using continuous profiling to inform code improvements.

Optimizing Your Programs with Continuous Profiling

How we re-architected Parca Agent to profile every process on the host and made it more scalable in the process.

System-wide profiling in Parca Agent

Polar Signals is sponsoring open-source developers.

Sponsoring the Open-Source Community

We'll talk about what is HTTP 103 (Early Hints), how it can help speed up the web.

For a change, let's talk web performance: HTTP 103 (Early Hints)

FrostDB is an embedded columnar database built for observability data. This article will explore FrostDB's potential by using it in Prometheus.

Winter is coming - FrostDB in Prometheus

Parca is our open source, always-on eBPF-based continuous profiler. In this article, we'll learn some of the concepts behind Parca and use it to profile our system.

Introduction to Parca - Part 1

gRPC is a new protocol for building a backend for our frontend. In this article, we'll learn how we use gRPC and gRPC-web to build our frontend at Polar Signals.

How we use gRPC to build our frontend at Polar Signals

We made a mistake in naming our open-source in-memory columnar database.

Naming is hard: ArcticDB is now FrostDB (+updates!)

How training Brazilian Jiu Jitsu has helped me become a better programmer

What BJJ taught me about Software Development

Attending KubeCon EU 2022 in Valencia, Spain.

Polar Signals at KubeCon EU 2022

Continuous profiling needed a columnar database, so we built it.

Introducing arcticDB: A database for Observability

We will learn how to profile your Next.js/Node.js apps with Parca and how it helps you to go back in time and identify the root cause for the irregular bottlenecks in your application's performance.

Profiling Next.js apps with Parca

In a previous post we detailed how we build our APIs here at Polar Signals. But we're going back to basics, in this tutorial we'll teach you how to build a small App from API to implementation.

How we build our APIs: From Scratch

Lerna and Yarn Workspaces gives us the ability to build and publish libraries and apps in a single repository (a.k.a. Monorepo). In this article we'll see how to publish UI components from a monorepo.

Managing monorepos with Lerna and Yarn workspaces

How profilers and debuggers translate machine addresses to human-readable symbolic names

Fantastic Symbols and Where to Find Them - Part 2

Fantastic Symbols and Where to Find Them - Part 1

All about function inlining, how it helps us create performant software, how we can learn to work with it, and how it influences profiling.

Why Compiler Function Inlining Matters

Tools and strategies that we use to build APIs that we love at Polar Signals.

How we build our gRPC APIs at Polar Signals using buf

Thoughts on how early company culture shapes the future of it.

On Building Company Culture

Introducing Parca - Prometheus but for profiles

Building pprof profiles from scratch using Go

DIY pprof profiles using Go

Learn how to demystify Pprof Labels using Go

Demystifying Pprof Labels with Go

Continuously profiling your applications without having to make modifications!

Perfessor - Continuous Profiling Without Instrumentation

Our first major feature since our product launch: cumulative profiling metrics!

Continuous Profiler Cumulative Metrics

Our first publicly available service, to upload and share pprof profiles.

Introducing Polar Signals' Pprof Profile Sharing Service

Since founding Polar Signals, we have been heads down working on our first product, and today we are beyond excited to share it with you!

Our Blog

DIY pprof profiles using Go

Demystifying Pprof Labels with Go

Perfessor - Continuous Profiling Without Instrumentation

Continuous Profiler Cumulative Metrics

Introducing Polar Signals' Pprof Profile Sharing Service

Announcing Polar Signals Continuous Profiler