Troubleshooting guide

REVIEW DRAFT—CISCO CONFIDENTIAL
1-5
Cisco Wide Area Application Services Configuration Guide
OL-26579-01
Chapter 1 Introduction to Cisco WAAS
Benefits of Cisco WAAS
BIC TCP, page 1-6
Windows Scaling
Windows scaling allows the receiver of a TCP packet to advertise that its TCP receive window can
exceed 64 KB. The receive window size determines the amount of space that the receiver has available
for unacknowledged data. By default, TCP headers limit the receive window size to 64 KB, but Windows
scaling allows the TCP header to specify receive windows of up to 1 GB.
Windows scaling allows TCP endpoints to take advantage of available bandwidth in your network and
not be limited to the default window size specified in the TCP header.
For more information about Windows scaling, refer to RFC 1323.
TCP Initial Window Size Maximization
WAAS increases the upper bound limit for TCP’s initial window from one or two segments to two to four
segments (approximately 4 KB). Increasing TCP’s initial window size provides the following
advantages:
When the initial TCP window is only one segment, a receiver that uses delayed ACKs is forced to
wait for a timeout before generating an ACK response. With an initial window of at least two
segments, the receiver generates an ACK response after the second data segment arrives, eliminating
the wait on the timeout.
For connections that transmit only a small amount of data, a larger initial window reduces the
transmission time. For many e-mail (SMTP) and web page (HTTP) transfers that are less than 4 KB,
the larger initial window reduces the data transfer time to a single round trip time (RTT).
For connections that use large congestion windows, the larger initial window eliminates up to three
RTTs and a delayed ACK timeout during the initial slow-start phase.
For more information about this optimization feature, see RFC 3390.
Increased Buffering
Cisco WAAS enhances the buffering algorithm used by the TCP kernel so that WAEs can more
aggressively pull data from branch office clients and remote servers. This increased buffer helps the two
WAEs participating in the connection keep the link between them full, increasing link utilization.
Selective Acknowledgment
Selective Acknowledgement (SACK) is an efficient packet loss recovery and retransmission feature that
allows clients to recover from packet losses more quickly than the default recovery mechanism used by
TCP.
By default, TCP uses a cumulative acknowledgement scheme that forces the sender to either wait for a
roundtrip to learn if any packets were not received by the recipient or to unnecessarily retransmit
segments that may have been correctly received.
SACK allows the receiver to inform the sender about all segments that have arrived successfully, so the
sender only needs to retransmit the segments that have actually been lost.
For more information about SACK, see RFC 2018.