Change QUIC's delayed ack timer from 100ms to 25ms.

Change QUIC's delayed ack timer from 100ms to 25ms.

Rationale: This delay kicks in when the receiver is waiting for a second
data packet before sending an ack, and 100ms seems inordinately long for
this wait. The timer fires per-packet in low-bandwidth network paths
(BW < ~384 kbps), where more frequent acks helps with (i) ack clocking,
and (ii) better bw estimation for BBR.

Merge internal change: 72788368

R=jri@chromium.org, rch@chromium.org

[ramant (doing other things), 2014-08-12 21:22:15]

https://codereview.chromium.org/461183002/diff/1/net/quic/quic_connection_tes...
File net/quic/quic_connection_test.cc (right):

https://codereview.chromium.org/461183002/diff/1/net/quic/quic_connection_tes...
net/quic/quic_connection_test.cc:887: return
QuicTime::Delta::FromMilliseconds(kMaxDelayedAckTime);
hi jri,
  Used kMaxDelayedAckTime instead of hard coding 25. WDYT?

[Jana, 2014-08-13 01:28:38]

lgtm

Thanks, Raman!

https://codereview.chromium.org/461183002/diff/1/net/quic/quic_connection_tes...
File net/quic/quic_connection_test.cc (right):

https://codereview.chromium.org/461183002/diff/1/net/quic/quic_connection_tes...
net/quic/quic_connection_test.cc:887: return
QuicTime::Delta::FromMilliseconds(kMaxDelayedAckTime);
On 2014/08/12 21:22:15, ramant wrote:
> hi jri,
>   Used kMaxDelayedAckTime instead of hard coding 25. WDYT?

Sure, that sounds fine.

[Ryan Hamilton, 2014-08-13 16:28:24]

lgtm

[ramant (doing other things), 2014-08-14 19:59:30]

https://codereview.chromium.org/461183002/diff/1/net/quic/quic_connection_tes...
File net/quic/quic_connection_test.cc (right):

https://codereview.chromium.org/461183002/diff/1/net/quic/quic_connection_tes...
net/quic/quic_connection_test.cc:887: return
QuicTime::Delta::FromMilliseconds(kMaxDelayedAckTime);
On 2014/08/13 01:28:38, Jana wrote:
> On 2014/08/12 21:22:15, ramant wrote:
> > hi jri,
> >   Used kMaxDelayedAckTime instead of hard coding 25. WDYT?
> 
> Sure, that sounds fine.

Acknowledged.