You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We've been getting reports that marking a camera now takes over 30 seconds for some users in particularly camera-rich regions. This is unacceptable both in terms of visual feedback (see #46), but also in terms of server performance. Right now we compare the new potential camera to each camera in its zone-file, to check for overlaps with an existing camera. That works fine with a zone-file of a hundred or less cameras, but may be extremely inefficient for a densely packed area.
We need to improve our database schema and querying mechanism to get that performance way down from O(n). It was a shoddy db design to begin with, and now that corner-cutting has caught up with us.
The text was updated successfully, but these errors were encountered:
There is a small chance that this was not tied to the camera database design (#56), but was instead tied to the login database design (#57). If this is the case, then new users should be unaffected by this issue, and existing users will be unaffected after marking a single camera (which will migrate them to the new login database).
We need to do some testing to confirm whether this issue still exists.
We've been getting reports that marking a camera now takes over 30 seconds for some users in particularly camera-rich regions. This is unacceptable both in terms of visual feedback (see #46), but also in terms of server performance. Right now we compare the new potential camera to each camera in its zone-file, to check for overlaps with an existing camera. That works fine with a zone-file of a hundred or less cameras, but may be extremely inefficient for a densely packed area.
We need to improve our database schema and querying mechanism to get that performance way down from O(n). It was a shoddy db design to begin with, and now that corner-cutting has caught up with us.
The text was updated successfully, but these errors were encountered: