cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1969
Views
3
Helpful
7
Replies

GLBP - Need help in Understanding GLBP Load balancing technique

Hi Experts,

i'm not sure if this is the right forum to ask this question but could not resist myself from posting this because I could see some wonderful replies from this forum

I have 4 switches (R4 through R7) on which I'm running GLBP and R4 is the active AVG and R7 is the standby AVG ..

The output command is shown below:

on R4

=====
R4#show glbp brief

Interface   Grp  Fwd Pri State    Address         Active router   Standby router

Vl10        1    -   120 Active   10.10.10.1      local           10.10.10.7

Vl10        1    1   -   Listen   0007.b400.0101  10.10.10.7      -

Vl10        1    2   -   Listen   0007.b400.0102  10.10.10.5      -

Vl10        1    3   -   Active   0007.b400.0103  local           -

Vl10        1    4   -   Listen   0007.b400.0104  10.10.10.6      -

R4#

and on R7 :

=========

R7#sh glbp brief

Interface   Grp  Fwd Pri State    Address         Active router   Standby router

Vl10        1    -   115 Standby  10.10.10.1      10.10.10.4      local

Vl10        1    1   -   Active   0007.b400.0101  local           -

Vl10        1    2   -   Listen   0007.b400.0102  10.10.10.5      -

Vl10        1    3   -   Listen   0007.b400.0103  10.10.10.4      -

Vl10        1    4   -   Listen   0007.b400.0104  10.10.10.6      -

R7#

My understanding based on the above output is:

1. R4 is responsible for traffic sent to the MAC 0007.b400.0103 and it is listening to the other AVF's and will take over when something fails there.

Lets assume that AVG failes. now R7 will become Active AVG and my question is:

Does AVG take the sole responsiblity of handling the traffic sent to R4's Virtual MAC or can he delicate this to other AVF's . My observation is : AVG takes over..

Please correct me if I went wrong?


R7#sh glbp brief

Interface   Grp  Fwd Pri State    Address         Active router   Standby router

Vl10        1    -   115 Active   10.10.10.1      local           10.10.10.6

Vl10        1    1   -   Active   0007.b400.0101  local           -

Vl10        1    2   -   Listen   0007.b400.0102  10.10.10.5      -

Vl10        1    3   -   Active   0007.b400.0103  local           -

Vl10        1    4   -   Listen   0007.b400.0104  10.10.10.6      -

R7#

Thanks,

-Vijay Swaminathan.


7 Replies 7

John Blakley
VIP Alumni
VIP Alumni

Vijay,

My understanding is that the AVG is responsible for handing out virtual macs to clients when they send traffic. For example, your virtual address is 10.10.10.1. When a user arps for that address, the AVG is responsible for handing over one of the AVFs virtual mac addresses. So you could have hosts with different macs to the same address. Then the AVFs are responsible for forwarding the traffic for that mac address.

If an AVF fails, you can have another AVF take over for that virtual mac address for the time being, but the AVG will know that it failed. During this time period the AVG uses 2 timers. In this time the AVF listens for 2 mac addresses, and the AVG starts to cancel out the failed mac address by assigning a new mac to those that have the old mac address. Now they are sending traffic to a different forwarder. Once the timeout period is up, the AVF that was listening for 2 addresses is now only listening to the originally assigned mac address that the AVF was originally assigned.

The AVG in your case when it fails will be replaced by the standby AVG, much like HSRP or VRRP. Now this AVG becomes responsible for keeping track of the virtual mac addresses that are assigned to the AVF.

http://www.cisco.com/en/US/docs/ios/12_2t/12_2t15/feature/guide/ft_glbp.html

Look under virtual forwarder timers for the answer to your above question...

HTH,

John

HTH, John *** Please rate all useful posts ***

Giuseppe Larosa
Hall of Fame
Hall of Fame

Hello Vijay,

configuration guide for GLBP provide the following

http://www.cisco.com/en/US/docs/ios-xml/ios/ipapp_fhrp/configuration/12-4t/fhp-glbp.html#GUID-1A23B304-3BDF-45C4-AFFD-99B2AF5A858E

R7 becomes the new AVG and new standby AVG is elected among remaining nodes in GLBP group.

The AVF role that was performed by R4 might be delegated to another node depending on new AVG decisions.

Depending on load balancing method you can emulate multiple hosts with a single PC by simply changing its IP address at every attempt to access outside the common subnet.

I used this to test GLBP as the load balancing action is performed on ARP activity by AVG.

Hope to help

Giuseppe

Thanks Giuseppe and Jhon for taking time to respond to my question.

Sorry if I wasn't clear on my question.

My question was how does the new AVG (R7 in this case) decide on which router should be responsible for the traffic sent to the R4's VMAC? That part is still unclear to me.....

In this case, R7 took that responsibility. But based on the above discussion looks like it is not a mandate that AVG should take up that responsibility (just in this case it happened to ) but any router in that group can take up that responsiblity of taking the traffic sent to the broken VMAC.

-Vijay

Vijay,

The AVF is chosen by the weighting value:

R1:

glbp 1 weighting 100 (default)

R2:

glbp 1 weighting 150

R3:

glbp 1 weighting 200

If R1 is the AVG and has handed out R2's virtual mac in response to an arp request and R2 dies, then R3 would become the forwarder for R2's virtual mac address.

If you want to manipulate the forwarder election, you would change the defaults of the weighting values. Here's a link explaining it a little more in-depth:

http://fengnet.com/book/Cisco.IOS.Cookbook.2nd/I_0596527225_CHP_22_SECT_15.html

HTH,

John

HTH, John *** Please rate all useful posts ***

Hi John,

But I'm seeing something different here:

GLBP_Summary.png

I see this on R7:

R7#

*Mar  1 12:28:21.752: %DUAL-5-NBRCHANGE: IP-EIGRP(0) 100: Neighbor 10.10.10.4 (Vlan10) is down: holding time expired

R7#

*Mar  1 12:28:22.956: %SYS-5-CONFIG_I: Configured from console by console

*Mar  1 12:28:24.192: GLBP: Vl10 1 Standby: g/Active timer expired (10.10.10.4)

*Mar  1 12:28:24.192: GLBP: Vl10 1 Active router IP is local, was 10.10.10.4

*Mar  1 12:28:24.192: GLBP: Vl10 1 Standby router is unknown, was local

*Mar  1 12:28:24.196: GLBP: Vl10 1 Standby -> Active

*Mar  1 12:28:24.196: %GLBP-6-STATECHANGE: Vlan10 Grp 1 state Standby -> Active

*Mar  1 12:28:24.200: GLBP: Vl10 1.3 Listen: g/Active timer expired

*Mar  1 12:28:24.200: GLBP: Vl10 1.3 Listen -> Active

*Mar  1 12:28:24.200: %GLBP-6-FWDSTATECHANGE: Vlan10 Grp 1 Fwd 3 state Listen -> Active

R7#

*Mar  1 12:28:24.212: GLBP: Vl10 1.3 Active: j/Hello rcvd from lower pri Active router (135/10.10.10.6)

*Mar  1 12:28:24.216: GLBP: Vl10 1.3 Active: j/Hello rcvd from lower pri Active router (135/10.10.10.5)

R7#

*Mar  1 12:28:44.200: GLBP: Vl10 1 Standby router is 10.10.10.6

R7#

So my understandering from the debug messages is:

Messages starting with GLBP: VI10 1 means msgs pertaining to AVG

Messages starting with GLBP: VI10 1.3 means msgs pertaining to AVF

so for selecting backup AVF for the lost active AVF for that MAC, it uses the priority.

Please let me know if you think otherwise

-Vijay

Let me lab this up...you have me questioning stuff now

HTH, John *** Please rate all useful posts ***

Okay...here's what I see:

*Mar  1 00:17:47.319: GLBP: Fa0/1 Grp 1 Hello  in  10.25.0.3 VG Speak   pri 100 vIP 10.25.0.1 hello 3000, hold 10000 VF 2 Active  pri 167 vMAC 0007.b400.0102

*Mar  1 00:17:47.335: GLBP: Fa0/1 1 Standby router is unknown, was 10.25.0.5

*Mar  1 00:17:47.335: GLBP: Fa0/1 1.4 Listen: g/Active timer expired

*Mar  1 00:17:47.335: GLBP: Fa0/1 1.4 Listen -> Active

*Mar  1 00:17:47.335: %GLBP-6-FWDSTATECHANGE: FastEthernet0/1 Grp 1 Fwd 4 state Listen -> Active

*Mar  1 00:17:47.339: GLBP: Fa0/1 Grp 1 Hello  out 10.25.0.2 VF 4 Active  pri 135 vMAC 0007.b400.0104

*Mar  1 00:17:47.343: GLBP: Fa0/1 Grp 1 Hello  in  10.25.0.4 VG Speak   pri 100 vIP 10.25.0.1 hello 3000, hold 10000 VF 3 Active  pri 167 vMAC 0007.b400.0103

*Mar  1 00:17:47.371: GLBP: Fa0/1 Grp 1 Hello  in  10.25.0.4 VF 4 Active  pri 135 vMAC 0007.b400.0104

*Mar  1 00:17:47.371: GLBP: Fa0/1 1.4 Active: i/Hello rcvd from higher pri Active router (135/10.25.0.4)

*Mar  1 00:17:47.371: GLBP: Fa0/1 1.4 Active -> Listen

*Mar  1 00:17:47.371: %GLBP-6-FWDSTATECHANGE: FastEthernet0/1 Grp 1 Fwd 4 state Active -> Listen

*Mar  1 00:17:47.371: GLBP: Fa0/1 API MAC address update

HTH, John *** Please rate all useful posts ***
Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Review Cisco Networking products for a $25 gift card