Skip to content

Explore

Reading paths

Four routes through the observatory, from a storage concept to its evidence and underlying data.

Reliability and SMART

What does a warning or failure rate actually tell us?

  1. Start with the hard disk
  2. Distinguish a rate from a probability
  3. Understand SMART's blind spots
  4. Keep missing telemetry separate from zero
  5. Inspect the AFR definition
  6. Inspect the drive-day denominators and source data

Explore this topic

Backup and recovery

Can the system recover the state that matters?

  1. Start with the backup definition
  2. Separate snapshots, replicas and backups
  3. Distinguish verification from a restore test
  4. Understand failed persistence operations
  5. Define the recovery point objective
  6. Explore a synthetic interrupted-write state model

Explore this topic

Measure storage performance

Which operations, timestamps and populations are being counted?

  1. Separate the protocol from the device
  2. Examine workload arrivals and latency
  3. Check whether the device reached steady state
  4. Reproduce the synthetic latency example
  5. Calculate an ideal transfer time

Explore this topic

Inside flash storage

How do cells, controllers and workloads interact?

  1. Start with the solid-state drive
  2. Understand cells and threshold states
  3. Follow erase blocks and cleaning costs
  4. Examine interrupted programming
  5. Reproduce the erase-block arithmetic
  6. Convert TBW and DWPD ratings

Explore this topic

Find a specific question

Search all published content, browse analyses, or follow the historical timeline.