How to optimize performance when using acls, How to optimize performance when using acls -33 – Avaya P580 User Manual

Page 319

Advertising
background image

Avaya P550R, P580, P880, and P882 Multiservice Switch User Guide, v5.3.1

9-33

Configuring IP Routing

simplest case, such as no ACL present, a Flow is identified
only by the destination address of the packet, for a high rate
of L3FE re-use. In the most complex case, such as an ACL
that specifies the entire 5-tuple, the granularity of
identification will cause a very low rate of re-use.

FORE (Fast Out-of-Band Routing Engine): When an 80-
series Ethernet media module is not licensed for routing, the
media modules forward all routed packets to the supervisor.
The F-Chip on the supervisor module provides fast, direct
forwarding of L3 packets to their destination ports.

Hash Mode: The F-chip uses a 12-bit key to initially locate
available memory for Flows. In DA-only mode (default) it
builds the key from a sample of the Flow’s destination
address (DA). In SA-DA mode, it uses both the source and
destination addresses. The F-chip automatically selects which
12 bits to use, and when to re-hash with a different key.

Ingress F-chip: The F-chip that receives a packet from an
external source.

Layer 3 Forwarding Entry (L3FE): When a packet arrives
on a media module, the F-Chip needs to know where to
forward that packet. The L3 forwarding cache is a list that
identifies the path taken by a data packet through the switch.
Each entry in this list is a L3FE. Each entry is identified by a
combination of the Flow’s 5-tuple and the corresponding
ACL rule. If the F-chip does not find a match in the L3
forwarding cache, the packet is sent Slow Path to the CPU for
processing and forwarding. Once the CPU has determined
the destination, it updates the L3 forwarding cache on the F-
chips with the L3FE. Once updated, the F-chip can forward
future packets via Fast Path.

SA: Source IP Address.

Slow Path: When an ingress F-chip does not recognize a
packet compared to its cache of known Flows, the packet is
forwarded to the CPU to determine proper destination and
ACL Rule assignment. See Layer 3 Forwarding Entry.

How to Optimize Performance when using ACLs

There are several techniques to optimize performance. They are
related and must be considered together.

Recognize Performance Issues

Evaluate System Performance

Advertising
This manual is related to the following products: