chore: Improve performance of TenantID

[cyriltovena]

What this PR does:

This improves performance in case there's is only 1 tenant which is apparently 99% of the time. The multi tenant case is kept unchanged.

❯ benchstat before.txt after.txt
name                   old time/op    new time/op    delta
TenantID/single-16       84.9ns ± 1%    19.4ns ± 3%   -77.16%  (p=0.008 n=5+5)
TenantID/single#01-16     180ns ± 5%     197ns ±13%    +9.21%  (p=0.032 n=5+5)

name                   old alloc/op   new alloc/op   delta
TenantID/single-16        40.0B ± 0%      0.0B       -100.00%  (p=0.008 n=5+5)
TenantID/single#01-16     72.0B ± 0%     72.0B ± 0%      ~     (all equal)

name                   old allocs/op  new allocs/op  delta
TenantID/single-16         2.00 ± 0%      0.00       -100.00%  (p=0.008 n=5+5)
TenantID/single#01-16      2.00 ± 0%      2.00 ± 0%      ~     (all equal)

Basically saves us 2 allocs per push/query requests.

Which issue(s) this PR fixes:

Fixes #

Checklist

• Tests updated
• CHANGELOG.md updated - the order of entries should be [CHANGE], [FEATURE], [ENHANCEMENT], [BUGFIX] Don't think I need a changelog entry for that.

[aknuds1]

I can see there's a performance regression of about 10% CPU time for the multi-tenant case versus main.

My suggestion is to refactor the string parsing in TenantIDs into a function tenantIDsFromString, that is also called from TenantID so you don't have to call user.ExtractOrgID twice in the multi-tenant case. When I try this, the perf regression goes away.

✗ benchstat perf.txt perf-arve.txt         
goos: linux
goarch: amd64
pkg: github.com/grafana/dskit/tenant
cpu: AMD Ryzen 9 3950X 16-Core Processor            
                   │  perf.txt   │           perf-arve.txt           │
                   │   sec/op    │   sec/op     vs base              │
TenantID/single-32   19.96n ± 9%   20.46n ± 4%       ~ (p=0.589 n=6)
TenantID/multi-32    320.5n ± 3%   290.6n ± 2%  -9.34% (p=0.002 n=6)
geomean              79.98n        77.10n       -3.60%

                   │   perf.txt   │           perf-arve.txt            │
                   │     B/op     │    B/op     vs base                │
TenantID/single-32   0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=6) ¹
TenantID/multi-32    48.00 ± 0%     48.00 ± 0%       ~ (p=1.000 n=6) ¹
geomean                         ²               +0.00%               ²
¹ all samples are equal
² summaries must be >0 to compute geomean

                   │   perf.txt   │           perf-arve.txt            │
                   │  allocs/op   │ allocs/op   vs base                │
TenantID/single-32   0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=6) ¹
TenantID/multi-32    1.000 ± 0%     1.000 ± 0%       ~ (p=1.000 n=6) ¹
geomean                         ²               +0.00%               ²
¹ all samples are equal
² summaries must be >0 to compute geomean

[cyriltovena]

good call

❯ benchstat before.txt after.txt
name                old time/op    new time/op    delta
TenantID/single-16    89.9ns ± 3%    19.0ns ± 1%   -78.85%  (p=0.008 n=5+5)
TenantID/multi-16      184ns ± 4%     183ns ± 2%      ~     (p=0.548 n=5+5)

name                old alloc/op   new alloc/op   delta
TenantID/single-16     40.0B ± 0%      0.0B       -100.00%  (p=0.008 n=5+5)
TenantID/multi-16      72.0B ± 0%     72.0B ± 0%      ~     (all equal)

name                old allocs/op  new allocs/op  delta
TenantID/single-16      2.00 ± 0%      0.00       -100.00%  (p=0.008 n=5+5)
TenantID/multi-16       2.00 ± 0%      2.00 ± 0%      ~     (all equal)

[aknuds1]

LGTM, thanks!

[aknuds1]

CI is failing consistently in TestMultipleClients :/ Could it be because of the change?

[cyriltovena]

Seems like a flake