The highest tagged major version is
README
¶
Substation
Substation is a toolkit for routing, normalizing, and enriching security event and audit logs.
At a Glance
Substation is inspired by data pipeline systems like Logstash and Fluentd, but is built for modern security teams:
- Extensible Data Processing: Build data processing pipeline systems and microservices using out-of-the-box applications and 100+ data transformation functions, or create your own written in Go.
- Route Data Across the Cloud: Conditionally route data to, from, and between AWS cloud services, including S3, Kinesis, SQS, and Lambda, or to any HTTP endpoint.
- Bring Your Own Schema: Format, normalize, and enrich event logs to comply with the Elastic Common Schema (ECS), Open Cybersecurity Schema Framework (OCSF), or any other schema.
- Unlimited Data Enrichment: Use external APIs to enrich event logs affordably and at scale with enterprise and threat intelligence, or build a microservice that reduces spend in expensive security APIs.
- No Servers, No Maintenance: Deploys as a serverless application in your AWS account, launches in minutes using Terraform, and requires no maintenance after deployment.
- Runs Almost Anywhere: Create applications that run on most platforms supported by Go and transform data consistently across laptops, servers, containers, and serverless functions.
- High Performance, Low Cost: Transform 100,000+ events per second while keeping cloud costs as low as a few cents per GB. Vendor solutions, like Cribl and Datadog, can cost up to 10x more.
All of these data pipeline and microservice systems, and many more, can be built with Substation:
Transforming Event Logs
Substation excels at formatting, normalizing, and enriching event logs. For example, Zeek connection logs can be transformed to comply with the Elastic Common Schema:
Original Event |
Transformed Event |
|---|---|
|
|
Routing Data
Substation can route data to several destinations from a single process and, unlike most other data pipeline systems, data transformation and routing are functionally equivalent -- this means that data can be transformed or routed in any order.
In this configuration, data is:
- Written to AWS S3
- Printed to stdout
- Conditionally dropped (filtered, removed)
- Sent to an HTTPS endpoint
// The input is a JSON array of objects, such as:
// [
// { "field1": "a", "field2": 1, "field3": true },
// { "field1": "b", "field2": 2, "field3": false },
// ...
// ]
local sub = import 'substation.libsonnet';
// This filters events based on the value of field3.
local is_false = sub.cnd.str.eq({ object: { source_key: 'field3' }, value: 'false' });
{
transforms: [
// Pre-transformed data is written to an object in AWS S3 for long-term storage.
sub.tf.send.aws.s3({ bucket_name: 'example-bucket-name' }),
// The JSON array is split into individual events that go through
// the remaining transforms. Each event is printed to stdout.
sub.tf.agg.from.array(),
sub.tf.send.stdout(),
// Events where field3 is false are removed from the pipeline.
sub.pattern.tf.conditional(condition=is_false, transform=sub.tf.util.drop()),
// The remaining events are sent to an HTTPS endpoint.
sub.tf.send.http.post({ url: 'https://example-http-endpoint.com' }),
],
}
Alternatively, the data can be conditionally routed to different destinations:
local sub = import 'substation.libsonnet';
{
transforms: [
// If field3 is false, then the event is sent to an HTTPS endpoint; otherwise,
// the event is written to an object in AWS S3.
sub.tf.meta.switch({ cases: [
{
condition: sub.cnd.any(sub.cnd.str.eq({ object: { source_key: 'field3' }, value: 'false' })),
transform: sub.tf.send.http.post({ url: 'https://example-http-endpoint.com' }),
},
{
transform: sub.tf.send.aws.s3({ bucket_name: 'example-bucket-name' }),
},
] }),
// The event is always available to any remaining transforms.
sub.tf.send.stdout(),
],
}
Configuring Applications
Substation applications run almost anywhere (laptops, servers, containers, serverless functions) and all transform functions behave identically regardless of where they are run. This makes it easy to develop configuration changes locally, validate them in a build (CI/CD) pipeline, and run integration tests in a staging environment before deploying to production.
Configurations are written in Jsonnet and can be expressed as functional code, simplifying version control and making it easy to build custom data processing libraries. For power users, configurations also have abbreviations that make them easier to write. Compare the configuration below to similar configurations for Logstash and Fluentd:
Substation |
Logstash |
Fluentd |
|---|---|---|
|
|
|
Deploying to AWS
Substation includes Terraform modules for securely deploying data pipelines and microservices in AWS. These modules are designed for ease of use, but are also flexible enough to support managing complex systems. This configuration deploys a data pipeline that is capable of receiving data from API Gateway and storing it in an S3 bucket:
resources.tf |
node.tf |
|---|---|
|
|
Getting Started
You can run Substation on:
Development
VS Code is the recommended development environment for Substation. The project includes a development container that should be used to develop and test the system. Refer to the development guide for more information.
Testing
The development container is used to test the system locally and in the cloud. If don't use VS Code, then you should run the development container from the command line:
git clone https://github.com/brexhq/substation.git && cd substation && \
docker build -t substation-dev .devcontainer/ && \
docker run -v $(pwd):/workspaces/substation/ -w /workspaces/substation -v /var/run/docker.sock:/var/run/docker.sock -it substation-dev
To try the system locally, run this from the examples directory:
sh .devcontainer/post_start.sh && \
cd examples && \
make -s quickstart
To try the system in the cloud, choose an AWS example to deploy:
sh .devcontainer/post_start.sh && \
cd examples && \
aws configure && \
make -s check && \
make -s build && \
make -s deploy EXAMPLE=terraform/aws/dynamodb/cdc
After testing is complete, the cloud deployment should be destroyed:
make -s destroy EXAMPLE=terraform/aws/dynamodb/cdc
We do not recommend managing cloud deployments from a local machine using the examples Makefile. Production deployments should use a CI/CD pipeline with a remote state backend, such as Terraform, to manage infrastructure.
Licensing
Substation and its associated code is released under the terms of the MIT License.
Documentation
¶
Overview ¶
Example (SubstationCustomConfig) ¶
package main
import (
"context"
"encoding/json"
"fmt"
"github.com/brexhq/substation"
)
// Custom applications should embed the Substation configuration and
// add additional configuration options.
type customConfig struct {
substation.Config
Auth struct {
Username string `json:"username"`
Password string `json:"password"`
} `json:"auth"`
}
// String returns an example string representation of the custom configuration.
func (c customConfig) String() string {
return fmt.Sprintf("%s:%s", c.Auth.Username, c.Auth.Password)
}
func main() {
// Substation applications rely on a context for cancellation and timeouts.
ctx := context.Background()
// Define and load the custom configuration. This config includes a username
// and password for authentication.
conf := []byte(`
{
"transforms":[
{"type":"object_copy","settings":{"object":{"source_key":"a","target_key":"c"}}},
{"type":"send_stdout"}
],
"auth":{
"username":"foo",
"password":"bar"
}
}
`)
cfg := customConfig{}
if err := json.Unmarshal(conf, &cfg); err != nil {
// Handle error.
panic(err)
}
// Create a new Substation instance from the embedded configuration.
sub, err := substation.New(ctx, cfg.Config)
if err != nil {
// Handle error.
panic(err)
}
// Print the Substation configuration.
fmt.Println(sub)
// Print the custom configuration.
fmt.Println(cfg)
}
Output: {"transforms":[{"type":"object_copy","settings":{"object":{"source_key":"a","target_key":"c"}}},{"type":"send_stdout","settings":null}]} foo:bar
Example (SubstationCustomTransforms) ¶
package main
import (
"context"
"encoding/json"
"github.com/brexhq/substation"
"github.com/brexhq/substation/config"
"github.com/brexhq/substation/message"
"github.com/brexhq/substation/transform"
)
func main() {
// Substation applications rely on a context for cancellation and timeouts.
ctx := context.Background()
// Define and load the configuration. This config includes a transform that
// is not part of the standard Substation package.
conf := []byte(`
{
"transforms":[
{"type":"utility_duplicate"},
{"type":"send_stdout"}
]
}
`)
cfg := substation.Config{}
if err := json.Unmarshal(conf, &cfg); err != nil {
// Handle error.
panic(err)
}
// Create a new Substation instance with a custom transform factory for loading
// the custom transform.
sub, err := substation.New(ctx, cfg, substation.WithTransformFactory(customFactory))
if err != nil {
// Handle error.
panic(err)
}
msg := []*message.Message{
message.New().SetData([]byte(`{"a":"b"}`)),
message.New().AsControl(),
}
// Transform the group of messages. In this example, results are not used.
if _, err := sub.Transform(ctx, msg...); err != nil {
// Handle error.
panic(err)
}
}
// customFactory is used in the custom transform example to load the custom transform.
func customFactory(ctx context.Context, cfg config.Config) (transform.Transformer, error) {
switch cfg.Type {
case "utility_duplicate":
return &utilityDuplicate{Count: 1}, nil
}
return transform.New(ctx, cfg)
}
// Duplicates a message.
type utilityDuplicate struct {
Count int `json:"count"`
}
func (t *utilityDuplicate) Transform(ctx context.Context, msg *message.Message) ([]*message.Message, error) {
if msg.IsControl() {
return []*message.Message{msg}, nil
}
output := []*message.Message{msg}
for i := 0; i < t.Count; i++ {
output = append(output, msg)
}
return output, nil
}
Output: {"a":"b"} {"a":"b"}
Index ¶
Examples ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func WithTransformFactory ¶
func WithTransformFactory(fac transform.Factory) func(*Substation)
WithTransformFactory implements a custom transform factory.
Types ¶
type Config ¶
type Config struct {
// Transforms contains a list of data transformatons that are executed.
Transforms []config.Config `json:"transforms"`
}
Config is the core configuration for the application. Custom applications should embed this and add additional configuration options.
type Substation ¶
type Substation struct {
// contains filtered or unexported fields
}
Substation provides access to data transformation functions.
Example ¶
package main
import (
"context"
"encoding/json"
"fmt"
"github.com/brexhq/substation"
"github.com/brexhq/substation/message"
)
func main() {
// Substation applications rely on a context for cancellation and timeouts.
ctx := context.Background()
// Define a configuration. For native Substation applications, this is managed by Jsonnet.
//
// This example copies an object's value and prints the data to stdout.
conf := []byte(`
{
"transforms":[
{"type":"object_copy","settings":{"object":{"source_key":"a","target_key":"c"}}},
{"type":"send_stdout"}
]
}
`)
cfg := substation.Config{}
if err := json.Unmarshal(conf, &cfg); err != nil {
// Handle error.
panic(err)
}
// Create a new Substation instance.
sub, err := substation.New(ctx, cfg)
if err != nil {
// Handle error.
panic(err)
}
// Print the Substation configuration.
fmt.Println(sub)
// Substation instances process data defined as a Message. Messages can be processed
// individually or in groups. This example processes multiple messages as a group.
msg := []*message.Message{
// The first message is a data message. Only data messages are transformed.
message.New().SetData([]byte(`{"a":"b"}`)),
// The second message is a ctrl message. ctrl messages flush the pipeline.
message.New().AsControl(),
}
// Transform the group of messages. In this example, results are not used.
if _, err := sub.Transform(ctx, msg...); err != nil {
// Handle error.
panic(err)
}
}
Output: {"transforms":[{"type":"object_copy","settings":{"object":{"source_key":"a","target_key":"c"}}},{"type":"send_stdout","settings":null}]} {"a":"b","c":"b"}
func New ¶
func New(ctx context.Context, cfg Config, opts ...func(*Substation)) (*Substation, error)
New returns a new Substation instance.
func (*Substation) String ¶
func (s *Substation) String() string
String returns a JSON representation of the configuration.
Source Files
Directories
¶
| Path | Synopsis |
|---|---|
|
cmd
|
|
|
development/benchmark/substation
Benchmarks the performance of Substation by sending a configurable number of events through the system and reporting the total time taken, the number of events sent, the amount of data sent, and the rate of events and data sent per second.
|
Benchmarks the performance of Substation by sending a configurable number of events through the system and reporting the total time taken, the number of events sent, the amount of data sent, and the rate of events and data sent per second. |
|
Package condition provides functions for evaluating data.
|
Package condition provides functions for evaluating data. |
|
Package config provides structures for building configurations.
|
Package config provides structures for building configurations. |
|
examples
|
|
|
internal
|
|
|
aws/appconfig
package appconfig provides functions for interacting with AWS AppConfig.
|
package appconfig provides functions for interacting with AWS AppConfig. |
|
aws/s3manager
package s3manager provides methods and functions for downloading and uploading objects in AWS S3.
|
package s3manager provides methods and functions for downloading and uploading objects in AWS S3. |
|
config
package config provides configuration types and functions for Substation.
|
package config provides configuration types and functions for Substation. |
|
file
package file provides functions that can be used to retrieve files from local and remote locations.
|
package file provides functions that can be used to retrieve files from local and remote locations. |
|
log
Package log wraps logrus and provides global logging only debug logging should be used in condition/, process/, and internal/ to reduce the likelihood of corrupting output for apps debug and info logging can be used in cmd/
|
Package log wraps logrus and provides global logging only debug logging should be used in condition/, process/, and internal/ to reduce the likelihood of corrupting output for apps debug and info logging can be used in cmd/ |
|
media
package media provides capabilities for inspecting the content of data and identifying its media (Multipurpose Internet Mail Extensions, MIME) type.
|
package media provides capabilities for inspecting the content of data and identifying its media (Multipurpose Internet Mail Extensions, MIME) type. |
|
secrets
Package secrets provides functions for retrieving local and remote secrets and interpolating them into configuration files.
|
Package secrets provides functions for retrieving local and remote secrets and interpolating them into configuration files. |
|
Package message provides functions for managing data used by conditions and transforms.
|
Package message provides functions for managing data used by conditions and transforms. |
|
Package transform provides functions for transforming messages.
|
Package transform provides functions for transforming messages. |
Click to show internal directories.
Click to hide internal directories.