/
Best Practices for Getting the Most from LHO

Best Practices for Getting the Most from LHO

Cloud data platforms are powerful but complex. Costs can grow quickly due to:

  • Over‑provisioned clusters

  • Inefficient autoscaling

  • Poorly distributed workloads

  • Underutilized compute resources

  • Long‑running or redundant jobs

  • Serverless misuse

LHO helps organizations understand these inefficiencies and fix them with clarity and confidence.

Do:

  • Prioritize your most expensive and frequently running jobs

  • Use fixed clusters for predictable workloads

  • Review KPIs after each change

  • Run A/B tests to validate cluster configuration changes

  • Reevaluate workloads quarterly

Don’t:

  • Rely solely on autoscaling defaults

  • Assume larger clusters will solve performance issues

  • Ignore driver resource waste

  • Apply serverless without evaluating total cost

  • Treat job cost increases as negative without checking data volume changes

  • Leave SQL warehouses auto stop settings overly long (causes hidden spend)

  • Leave All-Purpose Compute auto termination settings overly long (causes hidden spend)