1058962 - Simulate places usage on a test machine to measure performance changes related to favicon size increase.

Status: RESOLVED WONTFIX

(Toolkit :: Places, defect)

Description

[Bernardo Rittmeyer [:rittme]]

Changing the maximum favicon size (bug 798223) can have some performance impacts on the database. To evaluate that, we can simulate browser usage using a test machine and make performance comparisons between profiles.

This tests should allow us to check for performance impacts from database fragmentation.

MattN suggested the following procedure on bug 798223:
1) Create a test places.sqlite DB using data from telemetry and choosing median values for bookmarks/history. For profile A have all favicons be 1KB and for profile B have them all b 5KB.
2) Delete X% of the places randomly
3) Insert a similar amount of new places
4) Repeat 2 & 3 some times
5) Compare the performance of profiles A and B (on the same computer with the same procedures above) using the existing telemetry plus new telemetry probes.

Comment 1

[Marco Mucci [:MarcoM]]

Hi Bernardo, can you assign a point value for this bug.