Performance of Large Albums

[kiancross]

I've been trying to improve the performance of my Lychee instance. I have a large collection of photos, which I have broken up into albums of about 300 photos each.

For a particular album (334 photos), I have found that the /api/v2/Album?album_id=ABC call takes 962 ms and returns 964kB of content (when uncompressed).

Is this the expected API load time for an album of this size? I am not so much concerned about the response time itself, but the load it puts on the server: it drastically reduces the requests per second for an album of this size.

If this sort of performance is expected, are there any plans to improve large-album performance, such as limiting the amount of data sent or paginating it with lazy scroll, etc.?

I also wonder if some of the API responses, especially those expected to be large, could be designed in such a way that public cacheing is possible.

[ildyria]

Is this the expected API load time for an album of this size? I am not so much concerned about the response time itself, but the load it puts on the server: it drastically reduces the requests per second for an album of this size.

On my server I have the following:

• 371 Images = 875ms
• 38 Images = 401ms
• 0 Images, 3 sub albums = 113ms
• 1029 Images = 2009ms
• 0 Images, 26 sub albums = 387ms

If this sort of performance is expected, are there any plans to improve large-album performance, such as limiting the amount of data sent or paginating it with lazy scroll, etc.?

I do have some planning to separate the data of the album from the photos (which would be very convenient for pagination etc.)
I haven't had time to completely do it but I started implementing "infinite scrolling": #2578

There are some interesting problems though that arise from that. At the moment whenever a user scrolls down in the page, we record that position, when coming back to the album we restore it. However with infinite scrolling, managing this becomes a bit more tricky as if you are opening the page, you no longer have the full dom of images above...

I also wonder if some of the API responses, especially those expected to be large, could be designed in such a way that public cacheing is possible.

Literally what I have been working on for the last week: #2859
:)