JELAJAH SOC
YO

Engineering

Ingestion pipeline

Every SIEM, whatever the badge on it, moves data through the same six stages. Understand them once and you can operate, debug and capacity-plan any platform. Click a stage to see what it does and what the event looks like as it leaves.

Ingest rate

14.2k EPS

End-to-end latency

555 ms

collection to searchable

Parse fallthrough

0.9%

events hitting the generic decoder

Daily volume

620 GB

hot tier 71% used

Collection

Agents, syslog receivers, cloud pull

One collector is disconnected; two feeds are buffering locally.

Event as it leaves this stage

%ACME-FW-106023: DENY src=198.51.100.23:44120 dst=10.30.1.5:443 proto=TCP

Raw at collection

What an engineer tunes here

  • Agent groups
  • Syslog listeners
  • Cloud connectors
  • Local buffer size

Throughput

14.2k EPS

Added latency

40 ms

Health

Degraded

Where data silently disappearsfailure modes
  • Collector down: obvious once you look, invisible if nobody monitors event counts per source.
  • Decoder mismatch after a vendor format change: events arrive but land as unparsed text, so field queries return nothing.
  • Timezone drift at normalisation: events exist but sort into the wrong hour and vanish from your time window.
  • Queue backpressure: everything is “working”, just forty minutes late — fatal during a live incident.
  • Retention rollover: the hunt window you assumed was 90 days quietly became 30.
Capacity arithmetic you should be able to do out loud

Average event size × events per second × 86,400 gives raw daily volume. Multiply by your retention days and your replication factor to get storage. At 620 GB/day with 14 days hot, that is roughly 8.7 TB hot before replication — which is why filtering low-value events at collection is the cheapest performance work available.

Filtering without losing evidence

Drop volume at the edge only where you can defend the decision: verbose debug channels, health checks, duplicate feeds. Never drop authentication, process creation, or anything referenced by an existing detection. When in doubt, route it to a cheaper tier rather than deleting it.

Learn this capability

Hands-on labs that build the technique behind this workspace. Skills transfer to any mainstream SIEM.