ClickHouse engineering resources

What is MTTR? MTTD, MTBF, and the incident metrics that matter

MTTR is the average time from incident start to service restoration. Learn the formulas behind MTTR, MTTD, MTTA, and MTBF, what a good target looks like, and how observability shrinks each one.

Last updated: Jul 29, 2026

LLM inference latency: TTFT, tokens per second, and what to measure

LLM inference latency breaks down into TTFT, time per output token, throughput, and end-to-end latency. What each metric means, what moves it, and how to measure it with OpenTelemetry.

Last updated: Jul 29, 2026

LLM guardrails: what they are and how to run them in production

LLM guardrails are programmatic controls that validate model inputs and outputs to block prompt injection, PII leaks, and off-topic responses, and every verdict they emit is an event worth monitoring.

Last updated: Jul 29, 2026

The incident response process: from alert to postmortem

The incident response process is the structured sequence teams follow when production breaks: detect, declare, mobilize, mitigate, resolve, and learn, with severity levels, defined roles, and blameless postmortems.

Last updated: Jul 29, 2026

What are the four golden signals of monitoring?

The four golden signals (latency, traffic, errors, and saturation) are the minimum metric set the Google SRE book recommends for monitoring any user-facing system.

Last updated: Jul 29, 2026

Gauge vs counter: Prometheus metric types explained

A counter only increases and is read through rate(); a gauge moves up and down and is read directly. How the four Prometheus metric types differ, why counters reset, and how to pick between them.

Last updated: Jul 29, 2026

AI SRE agents: can AI run your on-call?

AI SRE agents are LLM-based systems that investigate production alerts by querying logs, metrics, and traces. What they do well today, where they still fail, and what they need underneath.

Last updated: Jul 29, 2026

How to engineer cost-efficient open source observability with ClickHouse (ClickStack) - 2026 technical playbook

A technical guide for engineers to build a cost-efficient observability stack. Learn how to use ZSTD codecs, Materialized Views, and Tiered Storage with ClickHouse to reduce observability data footprint by 10x.

Last updated: Jan 7, 2026

Build a dashboard in Python with ClickHouse and Streamlit

In this guide, you'll learn how to build a Python dashboard using ClickHouse and Streamlit. We'll create a real-world example that visualizes Bluesky social media data, walking through everything from basic setup to interactive visualizations. Perfect for data scientists and analysts who want to share their insights through custom dashboards.

Last updated: Jun 2, 2025

Instrumenting OpenAI with OpenTelemetry (OTel)

In this guide, we’ll learn how to instrument the OpenAI client with OpenTelemetry (OTel) so that we can generate and collect observability data about our LLM calls.

Last updated: Aug 1, 2025

Real-time data visualization

This guide is all about real-time data visualization. We'll explore how it differs from normal visualization, see some examples, and learn about the tools we can use.

Last updated: Apr 11, 2025

What is a columnar database?

A columnar database stores each column on disk separately to cut I/O and compress aggressively. Here is what that means in practice and when to use one.

Last updated: May 25, 2026

Best AlloyDB alternatives in 2026 for enterprise PostgreSQL workloads

AlloyDB alternatives compared for 2026: Aurora, RDS, Neon, Crunchy Bridge, and ClickHouse Managed Postgres with benchmarks, tradeoffs, and decision criteria.

ClickHouse concurrency: how to size for user-facing analytics

How to size ClickHouse for high-concurrency, user-facing analytics: turn active users into query load, benchmark under production-like conditions, and configure per-query limits, admission controls, workload scheduling, and replicas.

Last updated: Jul 2, 2026

When to denormalize, when to join: A ClickHouse guide

Denormalization and normalization are both valid analytical data-modeling strategies. A decision framework for choosing where to denormalize, where to join, and which ClickHouse primitives bridge the gap.

Last updated: Jun 26, 2026