This article is intended to supply information about the local-bias percentage on MC-LAG and LAG bundle interface as well as to determine if the same can be configurable as well as if there is a command to confirm the percentage on Juniper Networks switching platforms.
Customer needs to know local-bias percentage on MC-LAG and the command that can be run to confirm the same for an interface. The customer may have tested this on Juniper Networks MX Series routing platforms.
According to the below documentation Load Balancing, load balancing of network traffic between MC-LAG peers is 100 percent local bias. Load balancing of network traffic between multiple LAG members in a local MC-LAG node is achieved through a standard LAG hashing algorithm.
For more information on the hashing algorithm of LAG bundle load balancing, please visit the link Load Balancing and Ethernet Link Aggregation Overview
Upon testing local-bias configuration on QFX5210 in lab as well as on an MX960 routing platform. The observation is that the local-bias percentage can be configured on the MX960 but cannot be configured on the QFX5210. Based on the first document mentioned above, the local-bias is enabled by default with its default values for MG-LAG setups, although if you issue the command below in VTY on a switching platform such as EX Series QFX Series, you will see local-bias at 0%, because the percentage is not configurable on switching platforms, although it is configurable on MX Series routing platforms at hierarchy level [edit interfaces <interface-name> aggregated-ether-options load balancing local-bias]. The load balancing hierarchy does not exist on switching platforms.
Note that this is the only command that I was able to find to see the local-bias of an interface, and it is not an operational command nor a regular shell command. Also, the percentage will change only after configuring specific load balancing local-bias only applicable for MX Series routing platforms as explained above.
Checking local-bias in VTY:
First, you need to go shell mode with user root (user@switch> start shell user root), then you need to go VTY (user@switch% vty fpc#). At this point, you can run the command (FPC0(SWITCH vty)# show interface <LAG-interface>). See the example below:
user@switch:RE:0% vty fpc0 Switching platform (2200 Mhz Pentium processor, 639MB memory, 0KB flash) FPC0(SWITCH vty)# show interface ae1 Physical device ae1 (Index 641, context 0x0) SNMP Index 563 IPC sequence number initialized NO sequence number 0 Administration interface Device flags: (0x0000000000000001) Present Flags: (0x0000000000008000) Up SNMP-Traps JAM Capability: FALSE Capability Flags: Level 1 = 0x0000000000000020, Level 2 = 0x0000000000000000 IFD Specific Flags: 0x0 Interface redirects: Disabled Sparse DLCIs : Disabled Controller Interface: FALSE Locally managed: TRUE Shared Uplink: FALSE Transceiver channelized? FALSE SA multicast: FALSE MUP/MDOWN rcvd: TRUE Pad to minimum frame size: Disabled ifl count: 1 Parent IFD: 641 ae1 LB Accounting: Off Adaptive LB: Tolerance: 20% pps(1), bps(0) selection: 0 Flaps: Flap flag: Cleared, Marked ups: 2, Marked downs: 0 Timed flag: Cleared, Down dtime: 0, Up dtime: 0 Up/down timer: Stopped Append tstamp: 0+00:01:17.853, Yank tstamp: 0+00:01:17.860 Media: Type: Ethernet (3), Clocking: Internal (1), Link Flags: 0x0 Encapsulation: Ethernet (1), Physical select: Unspecified (0) Up hold time: 0 ms, Down hold time: 0 ms Alternative hold time alg.: Disabled Speed: 200000 Mbps, MTU: 9192 bytes, Header: 14, Link select: 0 CoS capability: flags:0x0000 hw max q: 0 usable max q: 10 STS: 0 to 0 PDH: 0, Timeslot Mask: 0x00000000 (ucode)Pop all label mode: 0 Optimize for aggregate FRR: 0 Share standby (vmx-only) : 0 Dampening half life: 0 max suppress 0 reuse level 0 suppress level 0 enable 0 suppress state 0 priv flags 0 Framing mode: Unspecified Load Balancing Flags: 0x0 Local-bias 0% Unidirectional: Unspecified
Load Balancing
Load Balancing and Ethernet Link Aggregation Overview