Last Modified: Sep 13, 2023
Affected Product(s):
BIG-IP LTM
Known Affected Versions:
13.1.0, 13.1.0.1, 13.1.0.2, 13.1.0.3, 13.1.0.4, 13.1.0.5, 13.1.0.6, 13.1.0.7, 13.1.0.8, 13.1.1, 13.1.1.2, 13.1.1.3, 13.1.1.4, 13.1.1.5, 13.1.3, 13.1.3.1, 14.0.0, 14.0.0.1, 14.0.0.2, 14.0.0.3, 14.0.0.4, 14.0.0.5, 14.0.1, 14.0.1.1, 14.1.0, 14.1.0.1, 14.1.0.2, 14.1.0.3, 14.1.0.5, 14.1.0.6, 14.1.2, 14.1.2.1, 14.1.2.2, 15.0.0, 15.0.1, 15.0.1.1, 15.0.1.2
Fixed In:
15.1.0, 15.0.1.3, 14.1.2.3, 13.1.3.2, 12.1.5.1
Opened: Apr 04, 2019 Severity: 3-Major
When Appropriate Byte Counting is enabled (the default), TCP's congestion window increases slower in slow-start when the data receiver sends stretch ACKs.
ABC limits the increase of congestion window by 2*MSS bytes per ACK. TCP's congestion window is increased slower in slow-start, which may lead to longer transfer times.
-- TCP data sender receives stretch ACKs (ACKs that acknowledges more than 2*MSS bytes of data). -- Appropriate Byte Counting (ABC) is enabled in slow-start.
There is no workaround at this time.
A new sys db (TM.TcpABCssLimit) is provided to set TCP's ABC limit (the default is 2*MSS) on increasing congestion window per ACK. With a larger limit (default is 2*MSS), TCP's congestion window increases faster in slow-start when stretch ACKs are received. If the data receiver sends regular ACKs/delayed ACKs, this setting has no impact.
There is a new db variable, TM.TcpABCssLimit for specifying TCP's ABC limit (the default is 2*MSS) on increasing congestion window per ACK. With a larger limit (default is 2*MSS), TCP's congestion window increases faster in slow-start when stretch ACKs are received. Note: If the data receiver sends regular ACKs/delayed ACKs, this setting has no impact.