Memory ledger: heap, pools, and retention

Quantify fragmentation and long-lived allocations that threaten week-long stability.

Cover visual for Memory ledger: heap, pools, and retention

Duration

9 business days

Format

Written brief

Investment

₩4,600,000

Platform

Mbed

Scope

We baseline allocator behavior, map long-lived objects, and pair mitigation paths with acceptance tests your QA team can own.

Included analyses

  • Allocator trace plan with nightly diff
  • Struct packing pass on hottest structs
  • Pool sizing spreadsheet with failure budgets
  • TLS and mbedTLS footprint snapshot
  • Retention RAM usage map
  • Fault injection script for OOM paths

Outcomes you can publish internally

  1. Heap delta table with owners
  2. Pool configuration candidates
  3. Regression checklist for memory PRs

Lead advisor

Portrait for Sora Kim

Sora Kim

Benchmark analyst translating measurements into release criteria.

FAQ

Do you modify TLS stacks?

We advise; cryptographic changes stay with your security reviewers.

Dynamic memory allowed?

We document risk if dynamic allocation remains; elimination is your choice.

What is excluded?

No cloud-side memory profiling unless separately scoped.

Experience notes

“Helped us see why the coap pool was oversized while TLS lived on the edge.”

— Leo