Skip to main content

18.11 Troubleshooting & Performance Labs

Lab 1 — Remote query suddenly becomes slow​

Check:

  1. source latency,
  2. remote-query history,
  3. statistics,
  4. predicates,
  5. join behavior,
  6. source concurrency,
  7. recent source changes.

Lab 2 — Replication is behind SLA​

Measure:

source change
  -> extraction
  -> transfer
  -> target commit
  -> availability

Find the slowest stage before changing architecture.

Lab 3 — Source database is overloaded​

Determine whether:

  • too many federated queries,
  • poor predicates,
  • high concurrency,
  • missing source indexes,
  • expensive joins

are responsible.

Consider replication if the workload is analytical and repeatable.

Lab 4 — Duplicate records after retry​

Investigate:

  • batch identity,
  • source keys,
  • retry semantics,
  • target merge behavior,
  • partial commits.

Lab 5 — Datatype mismatch​

Compare source and target:

  • precision,
  • scale,
  • timezone,
  • encoding,
  • nullability.

Lab 6 — API ingestion misses records​

Check:

  • pagination,
  • cursor handling,
  • rate limits,
  • retry behavior,
  • source mutation,
  • watermark logic.

Lab 7 — Cloud warehouse cost spikes​

Determine whether a federated design is causing repeated large scans.

Then compare:

  • source-side optimization,
  • filtered replication,
  • local persistence,
  • semantic redesign.

Lab 8 — Security works for admin but not consumer​

Trace:

identity
 -> role
 -> space
 -> object privilege
 -> data access control
 -> consumption

Lab 9 — Source becomes unavailable​

Define:

  • consumer behavior,
  • cached or persisted fallback,
  • alerting,
  • recovery,
  • reconciliation.

Lab 10 — Architect review​

For a 20-source enterprise integration landscape, identify:

  • federation candidates,
  • replication candidates,
  • push candidates,
  • sources requiring redesign,
  • sources requiring retirement.