interview-prep

Crisp answer: Logscale (formerly Humio) is a high-performance log management and observability platform built for real-time log ingestion and search at scale. Mimecast uses it as their central logging platform. CrowdStrike acquired Humio in 2021 and rebranded it as Falcon LogScale.

What Logscale does:

Ingests:  Structured and unstructured logs from any source
          (applications, infrastructure, security tools)
Stores:   Compressed, indexed for fast search
Searches: Real-time streaming search as logs arrive
Alerts:   Trigger on log patterns (e.g. error rate spike, specific event)
Dashboards: Visualise log-derived metrics

Key differentiator vs Elasticsearch/Splunk:

Humio/Logscale was built on a bucketed time-series storage model optimised
for log data specifically:

- Ingest at millions of events per second
- Compress logs 15-25x (stores raw data, not indexes in the traditional sense)
- Search runs against compressed data in parallel
- Much cheaper storage than Elasticsearch at equivalent retention
- Real-time: search includes logs arriving right now

LogScale query language (LQL):

Logscale uses a pipe-based query language:

# Find all errors in the last hour
level=ERROR | count()

# Find slow HTTP requests
http.method=GET http.duration>500 | avg(http.duration) | sort(avg)

# Top 10 error messages
level=ERROR | groupBy(message) | sort(count, limit=10)

# Alert when error rate exceeds threshold
level=ERROR | timechart(span=1m) | where count > 100

# Fields extraction from unstructured logs
/(?P<status>\d{3}) (?P<path>\/\S+)/ | groupBy(status)

How logs get into Logscale:

FluentD / Fluent Bit:   Kubernetes log collector → Logscale
Vector:                 Modern log router, replaces Fluentd in many stacks
Logstash:               Part of ELK stack, can forward to Logscale
Direct API:             Applications send directly via HTTP API
Elastic Filebeat:       Agent that ships log files

In a Kubernetes environment:

Pod logs (stdout/stderr)
  → Fluent Bit DaemonSet (runs on every node)
  → Logscale ingest API
  → Searchable in Logscale UI

Benefits over just kubectl logs:
  - Historical: logs from deleted pods are retained
  - Correlation: search across all pods and services simultaneously
  - Alerting: set up alerts on log patterns
  - Audit: immutable log retention for compliance

Logscale vs the alternatives:

Tool Strengths Weaknesses
Logscale Fast ingest, cheap storage, real-time Less ecosystem than ELK, CrowdStrike pricing
Elasticsearch Rich ecosystem, flexible Expensive at scale, complex to operate
Splunk Enterprise features, mature Very expensive licensing
Loki (Grafana) Cheap, integrates with Grafana Query language less powerful
CloudWatch Logs Built into AWS, no setup Expensive at high volume, limited search

What to say in the interview:

"Logscale is Mimecast's central log platform. I know it as formerly Humio, acquired by CrowdStrike. Its key advantage over Elasticsearch is the storage model: it stores compressed raw data and searches in parallel rather than maintaining large indexes, which makes it much cheaper per GB of logs at scale. In Kubernetes the standard pattern is a Fluent Bit DaemonSet on every node that ships pod logs to Logscale. I have not used Logscale directly but I have used Loki which follows a similar philosophy, and the LQL query language is a pipe-based model similar to Splunk SPL. I am comfortable picking it up quickly."


My notes