Tests were performed on the original, unoptimized Courses API. These tests are not meant to be comprehensive; they are just to confirm that the API does in fact need serious optimization work.
Tests were performed with a single user waiting 1000 ms between requests.
...
Details:
- Endpoint: /api/course_structure/v0/courses/<course_key>/blocks+navigation
- Number of users: 1
- User wait per request: 1000 ms
| Course key | # Requests made | Req/s | Median response time (ms) | 95% response time (ms) | (Link to X-Ray trace) |
|---|---|---|---|---|---|
| edX/DemoX.1/2014 | blocks | ||||
| navigation | |||||
| blocks+navigation | 343 | 0.57 | 690 | 1400 | https://rpm.newrelic.com/accounts/88178/key_transactions/14280/x_rays/1744 |
course-v1:edx+DemoX.1+LT_SPLIT | blocks | ||||
| navigation | |||||
| blocks+navigation | 206 | 0.34 | 1800 | 2700 | |
MITx/6.002x/2013_Spring | blocks | ||||
| navigation | |||||
| blocks+navigation | 35 | 0.06 | 16000 | 21000 | |
course-v1:MITx+6.002x_6x+1T2015 | blocks | ||||
| navigation | |||||
| blocks+navigation | 30 | 0.05 | 18000 | 22000 |
...
...
...
...