Security organisations are continuously challenged to collect more logs, from more devices; a problem that is typically solved with a Security Information and Event Management (SIEM) platform.

Many traditional SIEMs try to solve all problems at once; at Origin, we decided to break the traditional model into discrete components, and use a combination of the best free and open source software and cloud services to run our SIEM smarter and cheaper, while avoiding vendor lock-in.

1. Shipping and Parsing:
   • We use a combination of Elastic Beats and Logstash, with some​ cloud-native pipelines where they make sense, for things like CloudTrail or VPC flow logs.
2. Analytics:
   • We split off only the subset of logs we need for our day-to-day operations and alerting into Splunk.
   • We use Amazon Athena to query our historical logs directly from archive, or any sources that aren't in Splunk.​
3. Archive:
   • We compress and partition our logs in Logstash before storing them in S3 for long term retention at very low cost.​

This repo consists of:

• An AWS CDK app that will help with provisioning Fargate and the associated AWS components to run Logstash without servers.
• Two docker images and associated Logstash configuration files to get you started.

1. We run a separate Logstash pipeline for each log source we’re ingesting, and we run them as separate microservices on Fargate.
2. We push the events from these listener services into a central Kinesis data stream – which acts as a buffer.
3. Then, we pull events from the data stream in batches and process them in a processor service, which is also Logstash running on Fargate.
   • This service parses any unstructured events, typically from syslog sources, it partitions events by time and event attributes, and it compresses these partitioned batches before uploading them to S3.
   • This service is also responsible for filtering off a subset of the event stream to a Splunk Universal Forwarder (not included in this stack).

AWS pricing can vary significantly from region to region. You must review and understand the costs of the CloudFormation templates this stack outputs before you deploy them.

This will vary depending on how many services you deploy, what size and how many of each task you run, and which region you deploy to.

For general guidance only. Origin Security's own telemetry pipeline implementation costs around USD $800/month to run in the Sydney region and manages around 400,000,000 events/day with regular peaks exceeding 10,000 events/second.

Review the documentation for how to get started.

This code to help you build the required AWS infrastructure, and Logstash sample configuration files, is licensed under the MIT license; Logstash itself is not.

Refer to https://github.com/elastic/logstash/blob/master/LICENSE.txt for details on Logstash's license.