Assessments

Performance Baseline

A read-only review of how your workload actually performs: latency percentiles per critical path, where the bottlenecks live, and a tuning backlog ranked by user impact — measured from the metrics you already have, before anyone reaches for a bigger instance.

From $7,000 fixed Two weeks Read-only access

Measured, not guessed

  • Latency and throughput baseline — p50/p95/p99 per critical path from ALB, API Gateway, and CloudWatch metrics you already emit
  • Database performance — Performance Insights, slow-query patterns, connection-pool saturation, and the N+1s your ORM is hiding
  • Resource saturation — where CPU, memory, IOPS, and network actually run out, versus where instances are simply oversized
  • Scaling behavior — auto-scaling response under real traffic patterns, cold-start and warm-up costs, headroom before the next traffic doubling
  • Caching posture — what is cached, what should be, and the hot read paths a cache would take off the database
  • Load-test execution — up to two iterations with AWS Distributed Load Testing, deployed in your account from our template by your team, so our access stays read-only while the numbers get made under real load

The mechanics

Read-only access to CloudWatch, Performance Insights, and your APM if you have one. Load tests run from AWS Distributed Load Testing deployed in your account by your team from our template — our access stays read-only throughout. One kickoff, one engineering interview, one walkthrough.

From $7,000 fixed: one workload, up to five test scripts, two load-test iterations. Additional iterations $1,500 each. Exact price confirmed in the scoping call.

Deliverables, not impressions

Performance baseline document Current p50/p95/p99 per critical path, documented — the before picture every future optimization and incident review gets measured against.
Bottleneck findings register Findings scored by user impact with the metric evidence named — CSV plus written report, same register discipline as every Uptempo Cloud assessment.
Tuning backlog Ranked recommendations with the expected effect stated per item — query fixes, caching, right-sizing, scaling policy changes — sequenced so cheap wins land first.
Load-test results Up to two test iterations against the workload — scripts, traffic profile, and the harness template are yours to keep and re-run.

Two weeks, start to walkthrough

  1. 1
    Days 1–2 — access and baseline

    Read-only access live; metrics history pulled; critical paths identified with your team.

  2. 2
    Days 3–8 — analysis and load tests

    Each scope area worked against the metrics; test scripts designed and two iterations run from your account; one engineering interview for the patterns numbers can’t show.

  3. 3
    Days 9–10 — report and walkthrough

    Baseline, register, and tuning backlog delivered; 60-minute walkthrough; access revoked.

Book this one when

  • Users are reporting slowness and the team is debating opinions instead of reading percentiles
  • A traffic event is coming — launch, campaign, seasonal peak — and nobody knows the headroom number
  • The proposed fix is “bigger instances” and you would like evidence before paying for that forever

Performance problems that turn out to be architecture problems — single-AZ databases, missing queues, chatty service boundaries — graduate to the full Architecture Assessment — half this fee credits toward it at its standard rate within 90 days.

Frequently asked questions

How do the load tests work?

We design the test scripts and traffic profile, and your team deploys AWS Distributed Load Testing in your account from the template we provide — so tests run with your hands on the deploy button and our access stays read-only. Two iterations are included; tests target the environment you choose (staging strongly recommended for the first run), and additional iterations are $1,500 each.

How is this different from the Architecture Assessment?

Depth on one pillar. This takes Performance Efficiency alone and goes deeper — per-path percentiles, database internals, scaling behavior — in two weeks instead of four. If the trigger is “the app is slow,” start here; if it is “is the whole environment sound,” start with the full assessment.

What if our monitoring is too thin to baseline?

That is itself a finding — and a common one. The scoping call will tell you honestly whether there is enough signal to work with; if not, the first deliverable becomes the instrumentation plan that makes a baseline possible, and we scope the price to match.

What does it cost?

From $7,000, fixed — one workload, up to five test scripts, and two load-test iterations, confirmed in the scoping call. Additional test iterations are $1,500 each, and implementation help with the tuning backlog is available as a separate engagement.

Get the numbers before the traffic does

Thirty minutes to scope it, a written fixed price within 24 hours, and a start date you can plan around.

Book the Scoping Call

Prefer email? Send the details instead →