appender: add initial set of benches

[bryangarza]

This patch adds blocking and nonblocking benchmarks. This code is from
an old PR (#703) that was never merged, and now ported to TOT so that it
compiles.

Co-authored-by: Zeki Sherif [email protected]

[davidbarsky]

Ran the benchmarks locally:

synchronous/single_thread
                        time:   [260.34 ns 261.33 ns 262.32 ns]
Found 5 outliers among 100 measurements (5.00%)
  4 (4.00%) high mild
  1 (1.00%) high severe
synchronous/multiple_writers
                        time:   [434.49 ns 439.19 ns 443.70 ns]

non_blocking/single_thread
                        time:   [359.14 ns 361.87 ns 365.46 ns]
Found 11 outliers among 100 measurements (11.00%)
  2 (2.00%) low mild
  6 (6.00%) high mild
  3 (3.00%) high severe
non_blocking/multiple_writers
                        time:   [449.06 ns 453.83 ns 460.30 ns]
Found 16 outliers among 100 measurements (16.00%)
  1 (1.00%) high mild
  15 (15.00%) high severe

[davidbarsky]

Ran these benchmarks on a slower, but less noisy, VM. Under the theory that the writers here where benchmarking the allocator more than tracing-appender, I've made the writers de-facto no-ops; returning only the length of the buffer to the caller. There are the results:

synchronous/single_thread
                        time:   [615.81 ns 621.69 ns 627.45 ns]
                        change: [-12.844% -11.995% -11.171%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 1 outliers among 100 measurements (1.00%)
  1 (1.00%) high mild
synchronous/multiple_writers
                        time:   [723.88 ns 736.01 ns 750.46 ns]
                        change: [-36.285% -33.383% -30.748%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 12 outliers among 100 measurements (12.00%)
  6 (6.00%) high mild
  6 (6.00%) high severe

non_blocking/single_thread
                        time:   [916.12 ns 925.63 ns 935.11 ns]
                        change: [-5.6573% -4.8139% -4.0233%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 2 outliers among 100 measurements (2.00%)
  1 (1.00%) high mild
  1 (1.00%) high severe
non_blocking/multiple_writers
                        time:   [1.0901 us 1.1054 us 1.1226 us]
                        change: [-58.656% -35.543% -10.128%] (p = 0.18 > 0.05)
                        No change in performance detected.
Found 7 outliers among 100 measurements (7.00%)
  4 (4.00%) high mild
  3 (3.00%) high severe

(The prior run was the PR's existing contents.)

I'll flamegraph these benchmarks to confirm that it's allocator overhead, but I suspect that we'd want to change the writers to return only the buffer size to measure just tracing-appender.

[bryangarza]

Thanks. I need to look at why the Netlify build is failing...

[davidbarsky]

same reason it was failing in Tokio: tokio-rs/tokio#4699 (comment). I'll fix this.