Buz

What is Buz?

Buz is a system for multi-protocol event collection, validation, annotation, and delivery.

It ships as a single lightweight binary for deployment flexibility.

Toss Buz into an AWS lambda function, GCP Cloud Run service, or K8s pod and start collecting events in minutes.

Multiple Input Protocols

Buz supports multiple input protocols including:

• Snowplow Analytics
• Cloudevents
• Self-describing JSON
• Webhooks

It even hosts a pixel for use in constrained tracking environments.

SDK's are supported out of the box so you can point existing Snowplow Analytics or Cloudevents tracking directly to Buz and it will just work™.

Or point your in-house SDK to it using self-describing JSON.

Multiple (Simultaneous) Destinations

Buz supports two dozen different destinations including:

• Redpanda
• Postgres
• Kinesis Firehose
• Google Pub/Sub
• Splunk
• AWS EventBridge
• TimescaleDB
• NATS
• ...and many more.

You can send events to one or more destinations, so fanning them out to where they need to Bee is simple. As is using Buz to migrate from one destination system to another.

Jsonschema-based validation

Every incoming payload is validated in microseconds using JSON Schema.

If a payload doesn't conform to the associated schema, it is marked as such.

If a payload doesn't have an explicitly-associated schema (such as the case with arbitrary webhooks and pixels), payload contents are not validated. It is enveloped as arbitrary for downstream processing.

Onboard Schema Registry

Buz ships with a lightweight schema registry that supports multiple schema backends including:

• GCS
• S3
• Minio
• Postgres
• Mysql
• Mongodb
• Local filesystem
• ..and more

Schemas are cached locally once sourced from the configured backend. Cache ttl and maximum size are configurable bu have sane defaults.

Schemas are available via HTTP at /s/$PATH_TO_SCHEMA or /s/$SCHEMA_NAME, depending on the backend.

The onboard schema cache can be purged via a GET or POST to the /c/purge route.

Payload Enveloping

Each incoming payload is wrapped in a lightweight envelope.

This envelope appends a bit of metadata such as isValid, buzTimestamp, schema vendor, namespace, version, an event uuid, the associated protocol, etc. Metadata is then used to power payload routing and sharding as well as namespace-level statistics.

As an example of an arbitrary pixel event, wrapped in said envelope:

{
    "uuid": "1f9a7a20-8fa7-4179-a0c2-35a80783854a",
    "timestamp": "2023-05-03T02:50:59.464042Z",
    "buzTimestamp": "2023-05-03T02:50:59.464042Z",
    "buzVersion": "x.x.dev",
    "buzName": "buz-bootstrap",
    "buzEnv": "development",
    "protocol": "pixel",
    "schema": "io.silverton/buz/pixel/arbitrary/v1.0.json",
    "vendor": "io.silverton",
    "namespace": "buz.pixel.arbitrary",
    "version": "1.0",
    "isValid": true,
    "payload": {
        "msg": "hello",
        "subject": "zander"
    }
}

Why Buz?

It's lightweight

It minimizes the typical infrastructure footprint of collecting events from many different sources and allows for low-infrastructure, highly-flexible rollouts.

It's flexible

Buz doesn't care what your existing systems look like or what you want them to look like in the future.

It helps with the "now", and helps get your infrastructure to where you'd like it to be (without another migration).

It saves time and money

Buz aims to improve the lives of pipeline maintainers and drastically reduce long-term maintenance of event collection systems.

Roll it out fast, keep it going without much thought, and shut it off when it isn't doing anything.

Try it out

(No, you don't need to talk to anyone. Though we're relatively friendly and there's a Discord if you want to...)

You'll need go on your machine. But don't need to be a gopher.

Bootstrapping Buz

Clone:

$ git clone [email protected]:silverton-io/buz.git && cd buz

Bootstrap:

$ make bootstrap

Sending sample events

Events will be sent to two sinks by default - colorized envelopes will be sent to stdout and sent to buz_events.json or buz_invalid_events.json files.

POST a cloudevent

curl -X POST localhost:8080/cloudevents -H 'Content-Type:application/cloudevents+json' -d '{"dataschema":"io.silverton/buz/example/gettingStarted/v1.0.json", "data": {"userId": 10, "name": "you", "action": "didSomething"}}'

POST an arbitrary webhook

curl -X POST "localhost:8080/webhook" -H 'Content-Type:application/json' -d '{"arbitrary": "thing"}'

POST a named (schematized) webhook

curl -X POST "localhost:8080/webhook/io.silverton/buz/example/generic/sample/v1.0" -H 'Content-Type:application/json' -d '{"id": "10"}'

POST self-describing JSON

curl -X POST "localhost:8080/self-describing" -H 'Content-Type:application/json' -d '{"payload": {"schema": "io.silverton/buz/example/generic/sample/v1.0.json", "data": {"id": "10"}}}'

GET an arbitrary pixel

localhost:8080/pixel?msg=hello&subject=world

GET a named (schematized) pixel

localhost:8080/pixel/io.silverton/buz/example/generic/sample/v1.0?id=10

Buz plays nicely with others

Quickstart documentation for setting up a lightweight streaming stack with Buz, a sample ui, nginx, Redpanda, and Kowl can be found here.

Deploying Buz

Buz can be deployed in a variety of ways. We've included end-to-end (terraformed) samples for AWS and GCP:

• Buz on AWS Lambda
• Buz on GCP Cloud Run

Documentation

Full documentation can be found here.

🍻🐝