Is Bilis open source?

It is source-available under the Functional Source License (FSL-1.1-ALv2): you can read, run and modify it, self-hosting included, and each release converts to Apache 2.0 two years after it ships. It is not an OSI-approved licence until that conversion, so we do not call it open source.

Can I run Bilis and Datadog side by side?

Yes. An OpenTelemetry Collector can export the same logs, traces and metrics to both, so you can compare them on real traffic before moving anything.

Does Bilis replace all of Datadog?

No. It covers logs, traces and metrics for a team's own services. If you rely on Datadog for alerting, real-user monitoring, synthetics or security products, Bilis does not replace those today.

How much disk does self-hosting take?

Raw log volume × retention ÷ compression. Columnar storage commonly compresses logs 5–15×; the calculator works it out for your rate.

Bilis vs Datadog

A self-hosted alternative to Datadog for logs, traces and metrics

Datadog does far more than Bilis, and bills by the gigabyte for it. Bilis does less — logs, traces and metrics for your own services — on a server you own, with no per-GB line, and with an MCP server so your coding agent can read the same data you do. This page is the honest version of that trade.

01 The short answer

Choose Bilis when

your log bill grows faster than your traffic, and you sample or drop logs to keep it down;
your logs carry data that should not leave your infrastructure;
you want a coding agent reading production telemetry without shipping it to another vendor;
you already send OpenTelemetry, or are about to.

Stay on Datadog when

you need alerting, on-call and dashboards you build yourself today;
you use its other products — real-user monitoring, synthetics, security;
you run a large fleet that needs replication and horizontal scale;
nobody on the team wants to operate a server.

02 Side by side

What each one is

Aspect	Bilis	Datadog
Where it runs	On your own server, or the hosted Free plan at bilis.app.	Datadog's cloud only; the Agent runs on your hosts and ships to it.
How log pricing works	Self-hosted: whatever the server costs. No per-GB or per-event line.	Per GB ingested plus per million events indexed, by retention period.
Signals	Logs, traces and metrics.	Logs, APM traces, metrics, and much more: RUM, synthetics, security, profiling.
How data gets in	OpenTelemetry over HTTP (JSON or protobuf), a Monolog channel, a one-line Linux agent.	The Datadog Agent and its integrations; OpenTelemetry is accepted too.
Storage you can query	OpenTelemetry's own table layout in ClickHouse — readable with plain SQL.	A proprietary store, read through Datadog's UI and API.
AI agents	Built-in MCP server: read-only, OAuth with a consent screen, queries run on your box.	An MCP server too, reading data held in Datadog's cloud.
Alerting	Not yet — keep whatever pages you today.	Mature monitors, notifications and on-call integrations.
Dashboards	A built-in overview and curated host charts; no dashboard builder.	A full dashboard builder.
Scale	One node, no replication. Sized for one team's services.	Built for very large fleets.
Source	Source-available under FSL-1.1-ALv2; each release becomes Apache 2.0 after two years.	Closed source.

03 What logs cost

The same logs, priced two ways

At Datadog's published list price for Log Management — $0.10 per ingested GB and $1.70 per million events indexed for 15 days, billed annually, as checked on 9 October 2026 — against the disk the same retention takes self-hosted. 500-byte events; negotiated discounts and other retention periods change the left column.

Volume	Ingested a month	Datadog logs, a month	Bilis, disk for 15 days
50 events/s	65 GB	$227	3.2 GB
1,000 events/s	1.3 TB	$4,536	65 GB
10,000 events/s	13.0 TB	$45,360	648 GB

Self-hosted cost is the server, not the gigabytes — the disk column assumes a 10× compression ratio, which you should measure. Run your own numbers →

04 Honest limits

What you give up

No alerting yet It is on the way. Until it lands, keep whatever already pages you.
One node No replication and no cluster. Back the data up to object storage from day one.
One retention period Thirty days by default for logs, spans and metrics, table-wide rather than per project.
Token search Whole words, case-insensitive, over the log body — not substrings or regular expressions.
OTLP over HTTP only gRPC senders go through an OpenTelemetry Collector, which bridges that hop.

The full list on the features page →

05 Trying it

Run both before you move anything

01
Get an endpoint

Self-host Bilis on one server, or start on the hosted Free plan. Self-hosting has no plan at all and stays free.
02
Fan out

Add Bilis as a second exporter in your OpenTelemetry Collector. Datadog keeps receiving everything it did.
03
Compare on real traffic

Search the same incident in both, connect your coding agent over MCP, and decide with your own data.

Start free OTLP endpoints The MCP server

06 Questions

Is Bilis open source?: It is source-available under the Functional Source License (FSL-1.1-ALv2): you can read, run and modify it, self-hosting included, and each release converts to Apache 2.0 two years after it ships. It is not an OSI-approved licence until that conversion, so we do not call it open source.
Can I run Bilis and Datadog side by side?: Yes. An OpenTelemetry Collector can export the same logs, traces and metrics to both, so you can compare them on real traffic before moving anything.
Does Bilis replace all of Datadog?: No. It covers logs, traces and metrics for a team's own services. If you rely on Datadog for alerting, real-user monitoring, synthetics or security products, Bilis does not replace those today.
How much disk does self-hosting take?: Raw log volume × retention ÷ compression. Columnar storage commonly compresses logs 5–15×; the calculator works it out for your rate.

Datadog is a trademark of Datadog, Inc. This comparison is ours, made from public information as of October 2026; if something here is out of date, tell us and we will fix it.