Overview

Backends define where your logs live and how Sazabi queries them. Unlike data sources, which bring data into Sazabi, backends determine where Sazabi looks when Chat needs to answer your questions.

What is a backend?

A backend is the storage layer that Sazabi queries when you ask Chat a question. When you ask "why are errors elevated on the checkout service?", Sazabi routes that query to the appropriate backend, retrieves the relevant logs, and analyzes them to give you an answer.

Sazabi supports two types of backends:

• Sazabi backend: Logs stored in Sazabi's managed ClickHouse infrastructure
• External backends: Your existing observability platforms (Datadog, CloudWatch, etc.)

Sazabi backend vs external backends

Sazabi backend

When you send logs to Sazabi using a data source, they are stored in Sazabi's managed ClickHouse backend. This is the default storage for all projects and provides:

• Fast full-text search across all your logs
• Time-series queries and aggregations
• Automatic schema detection
• AI-optimized indexing

External backends

External backends let Sazabi query your existing observability platform directly. Instead of storing a copy of your data, Sazabi authenticates with your platform and fetches logs on demand when Chat needs them.

This is useful when:

• You have compliance requirements that prevent data duplication
• You want to keep using your existing observability tooling
• You have significant historical data you do not want to migrate

How backends work

When you ask Chat a question, Sazabi:

1. Determines the backend: Based on your project configuration, Sazabi identifies which backend(s) to query.
2. Authenticates: For external backends, Sazabi uses your stored credentials to authenticate with the platform.
3. Routes the query: Sazabi translates your natural language question into the appropriate query format for each backend.
4. Retrieves results: Logs are fetched from the backend and returned to Chat for analysis.
5. Analyzes and responds: Chat processes the results and provides you with an answer.

Data sources vs backends

It is important to understand the difference between data sources and backends:

DATA SOURCE (Sending Logs)
+-------------------+         +-------------------+
|  Your Platform    | ------> |     Sazabi        |
|  (e.g. AWS CW)    |  logs   |    Storage        |
+-------------------+         +-------------------+
Data flows FROM your platform TO Sazabi

BACKEND (Querying in Place)
+-------------------+         +-------------------+
|  Your Platform    | <------ |     Sazabi        |
|  (e.g. AWS CW)    |  query  |    Chat           |
+-------------------+         +-------------------+
Sazabi queries your platform WITHOUT moving data

Data sources send a copy of your logs to Sazabi. The data is stored in Sazabi's backend, giving you fast queries, configurable retention, and full AI analysis capabilities.

Backends let Sazabi query your existing observability platform directly. Your data stays where it is, and Sazabi fetches what it needs on demand.

Some platforms support both modes. For example, AWS CloudWatch can be configured as a data source (forwarding logs to Sazabi) or as a backend (Sazabi queries CloudWatch directly).

When to use each

Use the Sazabi backend when

• You want the fastest query performance
• You are starting fresh with observability
• You want unified storage across all your platforms
• You need AI-optimized indexing and search

Use external backends when

• You have compliance requirements around data residency
• You have existing tooling and workflows you want to preserve
• You have significant historical data in another platform
• You want to avoid data duplication costs

Next steps