Open jswatsch opened 2 months ago
Sharing initial results of caching when running things locally with Redis as the caching database. The below results are for the endpoint: GET /worksets/{{worksetId}}/metadata. I'm seeing a 55% reduction in the average response time. :-D
I will proceed with implementing this for different endpoints in the TORCHLITE backend. Please let me know if you have any questions or concerns. Thanks.
CC: @dkudeki, @gworthey
That’s great news, Sandeep — thanks!
On Aug 1, 2024, at 1:36 PM, Sandeep Puthanveetil Satheesan @.***> wrote:
Sharing initial results of caching when running things locally with Redis as the caching database. The below results are for the endpoint: GET /worksets/{{worksetId}}/metadata. I'm seeing a 55% reduction in the average response time. :-D
I will proceed with implementing this for different endpoints in the TORCHLITE backend. Please let me know if you have any questions or concerns. Thanks.
CC: @dkudekihttps://urldefense.com/v3/__https://github.com/dkudeki__;!!DZ3fjg!8Sto3zE_3bqOa3gMvNrex4dx-A67M2YBToatkfxaFtQpLinHCqsqdAJFX3sHSBLZjN39V8PNu0AytRgBIoAgLD19SQqc$, @gwortheyhttps://urldefense.com/v3/__https://github.com/gworthey__;!!DZ3fjg!8Sto3zE_3bqOa3gMvNrex4dx-A67M2YBToatkfxaFtQpLinHCqsqdAJFX3sHSBLZjN39V8PNu0AytRgBIoAgLLXvw45L$
— Reply to this email directly, view it on GitHubhttps://urldefense.com/v3/__https://github.com/htrc/torchlite-backend/issues/111*issuecomment-2263719700__;Iw!!DZ3fjg!8Sto3zE_3bqOa3gMvNrex4dx-A67M2YBToatkfxaFtQpLinHCqsqdAJFX3sHSBLZjN39V8PNu0AytRgBIoAgLNQpwhL2$, or unsubscribehttps://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/ABHCTSR725477THIBA7QHY3ZPJ54BAVCNFSM6AAAAABKXDVEXOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENRTG4YTSNZQGA__;!!DZ3fjg!8Sto3zE_3bqOa3gMvNrex4dx-A67M2YBToatkfxaFtQpLinHCqsqdAJFX3sHSBLZjN39V8PNu0AytRgBIoAgLOTHenet$. You are receiving this because you were mentioned.Message ID: @.***>
Adding more test results for future reference. This includes all GET requests in the TORCHLITE backend.
1. No cache (base case) 2. When cache is building 3. After cache building is complete, i.e., all requests are going through the cache.
@dkudeki is reviewing pull request as of 9/5/2024
@dkudeki added a patch for a bug related to internal API calls. The bug seems to be isolated to internal endpoint calls, not a broader problem with caching.
@sandeep can deploy it to the development stack
Working on a first draft implementation of caching with https://github.com/long2ice/fastapi-cache and I'm currently testing the \/worksets endpoint.