diff options
| author | Roman Arutyunyan <arut@nginx.com> | 2023-07-27 16:37:17 +0400 |
|---|---|---|
| committer | Roman Arutyunyan <arut@nginx.com> | 2023-07-27 16:37:17 +0400 |
| commit | 6e60e21ac09a88a17e08cb4a15ebfcd6634ba10b (patch) | |
| tree | 1b0cb3a393806680113afdf9f2941329231acb68 | |
| parent | bdea5b703ff6f6fcf98ac8dd4e1e9e5c9ad05017 (diff) | |
| download | nginx-6e60e21ac09a88a17e08cb4a15ebfcd6634ba10b.tar.gz nginx-6e60e21ac09a88a17e08cb4a15ebfcd6634ba10b.zip | |
QUIC: optimized ACK delay.
Previously ACK was not generated if max_ack_delay was not yet expired and the
number of unacknowledged ack-eliciting packets was less than two, as allowed by
RFC 9000 13.2.1-13.2.2. However this only makes sense to avoid sending ACK-only
packets, as explained by the RFC:
On the other hand, reducing the frequency of packets that carry only
acknowledgments reduces packet transmission and processing cost at both
endpoints.
Now ACK is delayed only if output frame queue is empty. Otherwise ACK is sent
immediately, which significantly improves QUIC performance with certain tests.
| -rw-r--r-- | src/event/quic/ngx_event_quic_ack.c | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/src/event/quic/ngx_event_quic_ack.c b/src/event/quic/ngx_event_quic_ack.c index f194abfff..865be2e6a 100644 --- a/src/event/quic/ngx_event_quic_ack.c +++ b/src/event/quic/ngx_event_quic_ack.c @@ -1171,7 +1171,8 @@ ngx_quic_generate_ack(ngx_connection_t *c, ngx_quic_send_ctx_t *ctx) delay = ngx_current_msec - ctx->ack_delay_start; qc = ngx_quic_get_connection(c); - if (ctx->send_ack < NGX_QUIC_MAX_ACK_GAP + if (ngx_queue_empty(&ctx->frames) + && ctx->send_ack < NGX_QUIC_MAX_ACK_GAP && delay < qc->tp.max_ack_delay) { if (!qc->push.timer_set && !qc->closing) { |