Firenook's 40x Firebase Emulator Claim Rests on One Server
Software / news
Firenook's 40x Firebase Emulator Claim Rests on One Server
The Apache 2.0 project's own page reports 5.1 ms median writes against 208.2 ms, from one application on one machine that left the official stack swapping.

Firenook, an Apache 2.0 replacement for the Firebase emulators, says its median Firestore write takes 5.1 ms against 208.2 ms for Google's emulator stack, and that it peaks at 2.11 GiB of memory against 22.57 GiB. The test behind those figures ran on one server with one application, and the project's own benchmark page says so.
The tool appeared on Hacker News as a Show HN and had 40 points and six comments when we read it on Nov. 6. Firenook's site says it is an independent open-source project, not made by Google, at version 0.2.1; the benchmarks were recorded on 0.2.0, the first release meant for public use. The site names no maker. The source is at firenook-dev/firenook on GitHub.
What it swaps in
Firenook reads an existing firebase.json, keeps the same ports and SDK settings, and replaces the command firebase emulators:start with firenook emulators:start. It installs with npm install --save-dev firenook and needs Node 22, 24 or 26 on macOS, Linux with glibc 2.28 or later, or Windows x64. It needs no Java; the official stack in the project's benchmark, firebase-tools 15.22.0, ran on Java 26. Keep firebase-tools installed for deploying.
Google's Local Emulator Suite page lists Cloud Firestore, Realtime Database, Cloud Storage for Firebase, Authentication, Firebase Hosting, and beta Cloud Functions, Pub/Sub and Extensions. Firenook covers Firestore, Authentication, Cloud Storage, Cloud Functions, Extensions, Pub/Sub, Cloud Tasks and Eventarc. It does not emulate Realtime Database, Hosting, App Hosting or Data Connect; a firebase.json that configures them still starts, with a warning. A team that tests against the Realtime Database emulator cannot move over.
The numbers, and who produced them
All of these are vendor-supplied: the project ran them and no outside party is named as having repeated them. The page says an independent audit recomputed the figures from raw samples but does not say who did it.
- Official emulator stack208.2 ms
- Firenook5.1 ms
Source: Firenook benchmarks page (vendor-run), accessed 2026-11-06
| Measure | Official stack | Firenook |
|---|---|---|
| Peak emulator memory | 22.57 GiB | 2.11 GiB |
| Restart with data kept | 30.8 s | 0.59 s |
| Listener update | 209 ms | 11 ms |
| Storage download, median / p99 | 2.12 / 4.31 ms | 2.30 / 5.01 ms |
The last row is the one the project reports against itself: Firenook was slower on Storage downloads. A 500-id query with no fields was a tie, at 97.8 ms against 98.3 ms.

The one-server caveat
The workload was a production web app with an API, a Functions codebase and a cache worker, holding 211,205 Firestore documents and 33,352 Storage objects totalling 6.7 GB. Each emulator ran for two hours of that traffic, once, on an AMD Ryzen 5 3600 with 64 GiB of memory under Ubuntu 24.04 and Node 24.20.0, on Oct. 21 and 22. The page's own warning: "One server and one application. Measure your own workload before relying on these numbers."
The 10x memory gap has a second reading. The official stack outgrew the server: it began with 7.7 GB already in swap and paged out another 3.1 GB, while Firenook paged out nothing. The page says a machine with more memory would probably narrow the gap. The page also excludes one browser journey that ran longer on Firenook, because the official emulator wakes a waiting cache worker early, and says application start times include the app's own builds. Only the emulator-only starts and stops isolate the emulators.
For comparison with another local stand-in, the MiniLake Databricks emulator accepts any token and says so up front; Firenook's page is similarly candid about its limits. Its speed numbers belong with the other vendor-run benchmarks we have checked: useful for deciding what to measure, not for deciding what to buy. The version to watch is the release after 0.2.1, since the project calls 0.2.0 its first for public use.
Sources
- 01Firenook
More in Software
- 01Zigzag, a 35 MB Mac Browser, Courts Arc Users 35 Days Before ShutdownThe WebKit-based browser imports Arc spaces and pinned tabs, but its README lists no sync, no Boosts and a partial Chrome extension API.
- 02Nimony 0.6.2 Sets a Year-End Goal for Nim 3 With 44 Days LeftThe compiler that will become Nim 3 is MIT-licensed and builds natively without LLVM, but its FAQ says it cannot yet compile Nim 2 code unchanged.
- 03Alloy CSS Runs Menus and Tabs Without JavaScript, but Dialogs Need Safari 26.2The 24 KB MIT stylesheet leans on popover and invoker commands, so its no-script promise has browser floors that differ by component.
- 04Windows 11 Telemetry Flags Which Coding Agents You Run, Researcher FindsXusheng Li traced a service-experiment feature in build 26200.9974 that matches executable names such as claude.exe and queues the result as optional diagnostic data.