SIEM Offloading
There is a trend towards a second SIEM, and it's not new. Benefits include cost savings, new analytical capabilities, higher visibility, improved detection in a modern engine. And most importantly, incremental deployability: you can add a new system without disrupting existing services.
When you realize that you need to deploy two SIEMs, you are basically taking the first step towards a distributed architecture. While it's possible to run the offloading engine centrally, this is the time to re-evaluate your strategy. How to comply best with data residency regulations? How do I break down silos? How can I support threat hunting and detection engineering?
SIEM offloading with a new engine does not mean you have to immediately adopt a fully decentralized architecture. You can also build your own lakehouse architecture with VAST, thanks to a standardized data plane via Apache Arrow. In fact, it makes sense to centralize heavy-duty analytics that require a lot of horse power. But you can also push a lot of front-line detection deep into the edge.
Using VAST in front of your SIEM has the following benefits:
- Reduced cost: VAST cuts your bill by absorbing the load of the heavy hitters while you can keep using the long tail of integrated data sources without disruption.
- Higher performance: VAST's system architecture has a strict separation of read and write path that scale independently, making it possible to operate the system under continuous inbound load. Compared to legacy SIEMs, VAST is a resource-efficient, embeddable telemetry engine that offers 10-100x ingestion bandwidth, and executes queries with interactive latencies.
- Reduce Lock-in: VAST stores all event data in an open, analytics-friendly format (Parquet) that makes it easy to BYO detection workloads.
- Easy compliance: VAST's powerful transforms allow you to perform fine-grained field-level modifications to anonymize, pseudonymize, or encrypt sensitive data. With compaction, you can specify retention periods (e.g., "anonymize URLs after 7 days") and define a multi-level roll-up strategy to age data gracefully.