Creating qos policies, Policy map – Brocade Communications Systems Brocate Ethernet Access Switch 6910 User Manual

Page 855

Advertising
background image

Brocade 6910 Ethernet Access Switch Configuration Guide

805

53-1002581-01

Creating QoS Policies

39

Creating QoS Policies

Use the Traffic > DiffServ (Configure Policy) page to create a policy map that can be attached to
multiple interfaces. A policy map is used to group one or more class map statements (

page 802

),

modify service tagging, and enforce bandwidth policing. A policy map can then be bound by a
service policy to one or more interfaces (

page 812

).

Configuring QoS policies requires several steps. A class map must first be configured which
indicates how to match the inbound packets according to an access list, a DSCP or IP Precedence
value, or a member of specific VLAN. A policy map is then configured which indicates the boundary
parameters used for monitoring inbound traffic, and the action to take for conforming and
non-conforming traffic. A policy map may contain one or more classes based on previously defined
class maps.

The class of service or per-hop behavior (i.e., the priority used for internal queue processing) can be
assigned to matching packets. In addition, the flow rate of inbound traffic can be monitored and
the response to conforming and non-conforming traffic based by one of three distinct policing
methods as described below.

Police Flow Meter – Defines the committed information rate (maximum throughput), committed
burst size (burst rate), and the action to take for conforming and non-conforming traffic.

Policing is based on a token bucket, where bucket depth (that is, the maximum burst before the
bucket overflows) is specified by the “burst” field (BC), and the average rate tokens are removed
from the bucket is specified by the “rate” option (CIR). Action may be taken for traffic conforming to
the maximum throughput, or exceeding the maximum throughput.

srTCM Police Meter – Defines an enforcer for classified traffic based on a single rate three color
meter scheme defined in RFC 2697. This metering policy monitors a traffic stream and processes
its packets according to the committed information rate (CIR, or maximum throughput), committed
burst size (BC, or burst rate), and excess burst size (BE). Action may taken for traffic conforming to
the maximum throughput, exceeding the maximum throughput, or exceeding the excess burst size.

The PHB label is composed of five bits, three bits for per-hop behavior, and two bits for the
color scheme used to control queue congestion. In addition to the actions defined by this
command to transmit, remark the DSCP service value, or drop a packet, the switch will also
mark the two color bits used to set the drop precedence of a packet for Random Early
Detection. A packet is marked green if it doesn't exceed the committed information rate and
committed burst size, yellow if it does exceed the committed information rate and committed
burst size, but not the excess burst size, and red otherwise.

The meter operates in one of two modes. In the color-blind mode, the meter assumes that the
packet stream is uncolored. In color-aware mode the meter assumes that some preceding
entity has pre-colored the incoming packet stream so that each packet is either green, yellow,
or red. The marker (re)colors an IP packet according to the results of the meter. The color is
coded in the DS field [RFC 2474] of the packet.

The behavior of the meter is specified in terms of its mode and two token buckets, C and E,
which both share the common rate CIR. The maximum size of the token bucket C is BC and the
maximum size of the token bucket E is BE.

The token buckets C and E are initially full, that is, the token count Tc(0) = BC and the token
count Te(0) = BE. Thereafter, the token counts Tc and Te are updated CIR times per second as
follows:

If Tc is less than BC, Tc is incremented by one, else

if Te is less then BE, Te is incremented by one, else

Advertising