Addendum

8
Flex Hash and Optimized Boot-Up
This chapter describes the Flex Hash and fast-boot enhancements and contains the following sections:
• Optimizing the Boot Time
• Flex Hash Capability Overview
• Preserving 802.1Q VLAN Tag Value for Lite Subinterfaces
Flex Hash Capability Overview
This functionality is supported on the S6000 platform.
With the introduction of various overlay technologies such as network virtualization using generic routing
encapsulation (NVGRE) segments and Routable Remote Direct Memory Access (RRDMA) over Converged
Ethernet (RRoCE), information related to a traffic flow is contained in the L4 header. The fields in the L2
and L3 headers are not sufficient to distinguish the flows. Therefore, the fields in the L4 header are
processed when hashing is performed for packets over LAG and ECMP links. The Flex Hash functionality
enables you to configure a packet search key and matches packets based on the search key. When a
packet matches the search key, two 16-bit hash fields are extracted from the start of the L4 header and
provided as inputs (bins 2 and 3) for RTAG7 hash computation. You must specify the offset of hash fields
from the start of the L4 header, which contains a flow identification field.
You can cause the system to include the fields present at the offsets that you define (from the start of the
L4 header) as a part of LAG and ECMP computation. Also, you can specify whether the IPv4 or IPv6
packets must be operated with the Flex Hash mechanism.
Keep the following points in mind when you configure the Flex Hash capability:
• A maximum of eight flex hash entries is supported.
• A maximum of 4 bytes can be extracted from the start of the L4 header.
• The offset range is 0 – 30 bytes from the start of the L4 header.
• Flex Hash uses the RTAG7 bins 2 and 3 (overlay bins). These bins must be enabled for Flex Hash to be
configured. These bins contain the source module and source port information. These bins are
disabled by default in releases of Dell Networking OS earlier than Release 9.3.0.0. The default behavior
of disabling of these bins occurs because of incorrect egress port information that would otherwise
be displayed in the output of the diagnostic show command of show ip flow.
• If you configure the Flex Hash mechanism by using the load-balance ingress-port enable and
the load-balance flexhash commands, the show ip flow and show port-channel-flow
commands are not operational. Flex Hash settings and these show commands that display the Layer 3
packets and Layer 2 packets forwarding and flows are mutually exclusive; only either of these
capabilities can be functional at a point in time.
This behavior occurs because the Flex Hash capability is disabled by default, which causes the proper
functioning of show ip flow and show port-channel-flow commands. For the Flex Hash
algorithm to work on the S6000 platform, you must enter the load-balance ingress-port
enable
command, which preempts the usability of the IP or Layer 2 trace flow functionalities. This
Flex Hash and Optimized Boot-Up
131