Load and stress test
Does the platform hold up at production volumes before go-live: 200-claim batches, large attachments and high record volumes under concurrent load.
Read only. You can read this screen. Only the Super admin changes it.
Planned. The plan is agreed. The first run is booked for 12 Oct 2026 on the staging copy of the production setup.
Running. Run LT-03 started at 09:15 and runs for 2 hours. Results below fill in as each scenario finishes.
Thresholds met. Run LT-03 on 5 Oct 2026: every scenario is inside its threshold and the error rate stays under 1%. The go-live gate is passed.
Thresholds missed. Run LT-03 on 5 Oct 2026: worklist search and attachment uploads are over their thresholds. The go-live gate stays closed until the bottlenecks are fixed and the run is repeated.
Concurrent users
150
peak in the plan
Claims in the data set
120,000
plus 80,000 patients
Largest batch
200 claims
10 at once
Run
Not run yetLT-03
5 Oct 2026, 2 hours
Results against the thresholds
No results yet. Response times, error rates and the verdict per scenario appear here after the first run.
| Scenario | p50 | p95 | p99 | p95 against threshold | Error rate | Verdict |
|---|---|---|---|---|---|---|
| Submit a 200-claim batch10 batches at once | 1.8 s | 3.9 s | 5.2 s |
p95 3.9 s of 5.0 s
|
0.2% | Within |
| Upload a 9 MB attachment60 uploads at once |
|
Running | ||||
| Search the worklist120,000 records, 150 users |
|
Waiting | ||||
| Start an eligibility check150 users |
|
Waiting |
| Scenario | p50 | p95 | p99 | p95 against threshold | Error rate | Verdict |
|---|---|---|---|---|---|---|
| Submit a 200-claim batch10 batches at once | 1.8 s | 3.9 s | 5.2 s |
p95 3.9 s of 5.0 s
|
0.2% | Within |
| Upload a 9 MB attachment60 uploads at once | 2.1 s | 4.4 s | 6 s |
p95 4.4 s of 6.0 s
|
0.4% | Within |
| Search the worklist120,000 records, 150 users | 0.31 s | 0.82 s | 1.4 s |
p95 0.82 s of 1.0 s
|
0.1% | Within |
| Start an eligibility check150 users | 0.6 s | 1.3 s | 2.1 s |
p95 1.3 s of 2.0 s
|
0.1% | Within |
| Scenario | p50 | p95 | p99 | p95 against threshold | Error rate | Verdict |
|---|---|---|---|---|---|---|
| Submit a 200-claim batch10 batches at once | 2.4 s | 4.6 s | 6.1 s |
p95 4.6 s of 5.0 s
|
0.4% | Within |
| Upload a 9 MB attachment60 uploads at once | 3 s | 7.8 s | 11.4 s |
p95 7.8 s of 6.0 s
|
2.3% | Over |
| Search the worklist120,000 records, 150 users | 0.52 s | 1.9 s | 3.4 s |
p95 1.9 s of 1.0 s
|
0.6% | Over |
| Start an eligibility check150 users | 0.7 s | 1.5 s | 2.4 s |
p95 1.5 s of 2.0 s
|
0.2% | Within |
p50 is the typical call, p95 the slowest 1 in 20, p99 the slowest 1 in 100. Response times in seconds.
Test plan
| Scenario | Target load | Duration | Agreed p95 threshold |
|---|---|---|---|
| 200-claim batchesSubmit and read the NPHIES result | 10 batches at once | 30 min | 5.0 s |
| Large attachmentsUpload and link to a request | 60 uploads of 9 MB at once | 30 min | 6.0 s |
| High record volumesWorklist search over 120,000 claims and 80,000 patients | 150 users searching | 45 min | 1.0 s |
| Mixed dayEligibility checks while the other scenarios run | 150 concurrent users | 2 hours | 2.0 s |
Agreed with the Tech Leader. Error rate threshold for every scenario: 1% or less. Basic functional checks are on Basic performance checks.
Bottleneck log
Nothing logged yet.
| Finding | Owner | Status | Next step |
|---|---|---|---|
| Attachment upload queued behind batch submitsSeen at 60 uploads at once | SASamir Al-Qadi | Remediated | Watching while the run finishes. |
| Finding | Owner | Status | Next step |
|---|---|---|---|
| Attachment upload queued behind batch submitsSeen at 60 uploads at once | SASamir Al-Qadi | Remediated | Separate upload queue deployed. Re-run LT-03 confirmed it. |
| Batch submit slows after 8 batches at oncep95 stays inside the threshold | NHNadia Haddad | Logged | Accepted for go-live. Review after the first month of use. |
| Finding | Owner | Status | Next step |
|---|---|---|---|
| Worklist search scans the whole claims tablep95 1.9 s at 120,000 records | SASamir Al-Qadi | Logged | Add an index and a result limit. Re-run before 19 Oct 2026. |
| Attachment upload fails above 8 MB under loadError rate 2.3% | SASamir Al-Qadi | Logged | Raise the gateway limit to 10 MB, then repeat the upload scenario. |
| Batch submit slows after 8 batches at oncep95 4.6 s, still inside | NHNadia Haddad | Remediated | Connection pool raised from 20 to 40. |
Every bottleneck is either remediated or logged with an owner and a date. Nothing is left without a decision.
Go-live gate
Response times inside the agreed thresholds
Error rate 1% or less in every scenario
Bottlenecks remediated or logged with an owner
Basic performance checks passed
Not started
Response times inside the agreed thresholds
Error rate 1% or less in every scenario
Bottlenecks remediated or logged with an owner
Basic performance checks passed
In progress
Response times inside the agreed thresholds
Error rate 1% or less in every scenario
Bottlenecks remediated or logged with an owner
Basic performance checks passed
Gate passed
Response times inside the agreed thresholds
Error rate 1% or less in every scenario
Bottlenecks remediated or logged with an owner
Basic performance checks passed
Gate not passed
The gate also needs the signed-off API scoping spec and a pilot client without platform errors on Pilot client.
Run the test
Runs against the staging copy, never against live patient data. A run cannot start while another is running.