Cisco Switching/Routing :: 3750X Port Channel Load Balancing Protocols
Jan 15, 2012
What I am attempting to achieve is to aggregate trunk ports out of a VMware server into a single logical connection to give as much bandwidth as possible, the switches are 3750X and are three stacked together with the server connections spread across the stack. What I am not sure about is if two port channel load balance protocols can happy co-exist on the switch, by default the switch is using MAC address load balancing and Vmware wants to use IP Source load balancing. As other trunks and channels exist on the switch I don't want to make a change that will affect the other live connections if changing this is a global setting and not local on the channel.
I have an inquiry about a configuration I deploy in a C2960 switch. I have configured a ether channel with 8 ports, the load balanced method is source mac address. The bundling protocol is LACP.
I have found the ether channel is not balanced as I expect. One of the eight interfaces is congested.
Possible to configure multichassis port-channel between a VSS and 3750X (Port-channel not in the same switch on the 3750X stack). I got it using LACP but I need Pagp to get VSL redundancy “dual-active detection pagp trust channel-group 1”. I am using the last VSS IOS version 15.0(1)SY. I can’t get the option “switch 1 preempt”, has it changed with other option?
I need to tear down an existing port-channel on a 3750X running c3750e-universalk9-mz.150-1.SE3.bin. This port channel is currently down down. It has three ports in it that will be added one each to three existing port-channels, I am assuming as long as the "channel-group" command is exactly the same as it is on the three existing port-channels I should be ok just adding the new port. One point to note is that the three existing port-channels all have three ports so this will be adding the fourth port to each port-channel. I know after reading that it is a best practice for load balancing to use either 2, 4, or 8 ports for a port-channel. Also what is the command to see all ports that are in a port channel?
I have a dual-homed fabric (Nexus 2248 dual attached to two Nexus 5020's via vPC). On this Nexus 2248 is a server that has a four port LACP etherchannel. The ports do not appear to be load balancing correctly. The output below shows the four ports in use and it clearly shows port e138/1/10 as getting the most use. When I use the "show port-channel load-balance forwarding-path..." command on either of the vPC switches for various source and destination IP's that use this link, it shows them correctly load-balancing across the four ports. But we do not see this when looking at stats on both the server side and the switch side.
**************** Config info below. This is a vPC pair and the port configs are identical on both switches so I'm only showing the configs for one switch to keep it simple.
dc5020-3g# sh port-channel load-balance Port Channel Load-Balancing Configuration: System: source-dest-ip Port Channel Load-Balancing Addresses Used Per-Protocol: Non-IP: source-dest-mac IP: source-dest-ip source-dest-mac
I have two stacks 3750X on two different sites with two links L_2_L, and I want to configure the port channel to aggregate the two links.
Site A Site B 3750X -A1 --------------------------------------( )--------------------------------------- 3750X -B1 ( L-2-L ) 3750X -A2 --------------------------------------( )--------------------------------------- 3750X -B2
Below the configuration that I have put the two stacks.
site A interface Port-channel5 description Etherchannel group entre le stack 3750X-A et Switch Lan_2_Lan switchport trunk encapsulation dot1q switchport trunk allowed vlan 11,12,999 switchport mode trunk switchport nonegotiate speed 100
But the problem is only one link is Bundeled in channel group, see below
Group Port-channel Protocol Ports ------+-------------+-----------+----------------------------------------------- 5 Po5(SU) LACP Gi1/0/15(I) Gi2/0/15(P)
i have a VSS core that has 4 downstream links to 4 stacked switches. I connected each of the 4 links to different switch in the switch stack. I then created a port-channel that combines all the links from each switch. Is there any issues associated with this setup?
We have Nexus 7K on production. 7K chasis is not load balancing with non-cisco devices with etherchannel or LACP..I have tried all load balancing algorithms but in vain. [code]
I have conneccted a server with LACP on Nexus extender. I am starting different file copy from diffferent sources to this server, it does not load balance.
We have to cisco WS-C4900M with Cisco IOS Software, Catalyst 4500 L3 Switch Software (cat4500e-IPBASE-M), Version 12.2(53)SG5, RELEASE SOFTWARE (fc1).We have four gigabit link connected between those two switches.We have create a LACP port channel with those four ports on both switches. Ether-channel is up and running and defined with a load-balancing method of src-dst-ip.But when we test the load-balancing, it's not using the src-dst-ip rule with the XOR: [code]
I have two 3750Xs for stack, and i want connect a 2960 switch on this stack with port-channel.I connect 2960 G0/1 to 3750X G1/0/24, and 2960 G0/2 to 3750X G2/0/24. [code] i try to config spanning-tree etherchannel guard misconfig, but no use finally, i have to use "no errdisable detect cause all"..i check the cisco.com, and find a information: for this case, channel-group can use mode desirble.but when i try to change the channel-group in 3750X stack, it shows me that: "with pagp enable, desirble mode must config in the same switch".i try to use mode desirble mode in G1/0/23 and G1/0/24 in 3750X stack, it still can not work.i want to know how to solve this problem? not only do "no errdisable detect cause all".
I would like configure a CSS content, that uses the sorry service principal in an advanced way.
I am familiar with the primary Sorry Server command and see that the CSS would send all connections to the named service that is configured as the primary Sorry Server.
What I would like to do is to configure the CSS, so that once it’s decided it’s in a “sorry” state (all the services that are configured with “add service” are down) that it load balances to a different set of services.
To explain what I’ve been trying to do in the form of configuration on the CSS, I’ve pasted some pretend config below.
Connections come into IP address 1.1.1.1, which normally get load balanced between 9.1.1.1, 9.1.1.2 and 9.1.1.3.
If 9.1.1.1, 9.1.1.2 and 9.1.1.3 are all down, the sorry service is used and the CSS starts passing traffic to 1.1.2.1, which I want it to load balance between 9.1.2.1, 9.1.2.2 and 9.1.2.3.
The order that I have applied the config, is different to the below, as I set out to configure in this order: secondary services, secondary content, sorry service, primary services, primary content.
The order of the config below is different, because I wanted it in the order that the traffic flows and the CSS won’t take the config in that order!
The wall I have ran into, is that when I try to create the service I have named “Sorry Service”, I get the following error:
%% Service IP Address conflicts with a local I/F, VIP, mg mt route.
The best option for load balancing between 2 X Cisco nexus 5548UP switches located at one site and connecting to 2 X Cisco nexus 5548UP switches located at another site.
The sites are connected via a 1GB fibre connection. I am unable to use GLBP until GLBP is supported in further software releases.
I have probem with symmetric load balancig, in case when both ends of ether channel are on the sam switch (we are using VLAN translation).We need to create L2 port channel with both ends on same switch (Cisco WS-C4500X-24X-ES), for example:Po1 – Gi0/1, Gi0/3 (one end of port channel )Po2 – Gi0/2, Gi0/4 (other end of port channel)On ports in Po2 we will configure VLAN mapping.My question is what is the best ether-channel load-balancing scheme with wich we can accomplish full symmetry in both directions? For example, if traffic in one direction goes through Gi0/1 (member of Po1), in other direction also must go through Gi0/1. This is required because we need to connect four appliances for DPI (they are full L2 transparent) and traffic through each appliance need to be symmetric.
I can set-up src-ip, dst-ip, src-dst-ip etc. load balancing, but, actually I need src-ip on Po1 and dst-ip on Po2. Is there any way to set up different load balancing mechanism for different ether channel on same switch (4500X).
If the load balancing is set to src-dst-ip, will a layer 2 switch forward based on that information? Particularly talking about a 6500, with trunk interfaces, since those packets never go to the layer 3 engine, will the load-balancing work as intended?
Right now I have 2 default routes load balancing 100MB internet links. This is on my 2 6509's.
ip route 0.0.0.0 0.0.0.0 10.47.2.1 (FWSM) ip route 0.0.0.0 0.0.0.0 10.47.2.250 (5510)
Is there anyway to make the first default route take more of the traffic, like 60/40 or 70/30?Any program that I could use to see top users going through the FWSM?
I have the requirement to provide a Cisco Router with 3 x ADSL lines (768k) to increase the internet speed.PPP multilink is not supported from the ISP.
Is it possbile to distribute the traffic between this three ADSL lines?How can I configure this?
I have the following hardware configuration:
1 x CISCO1921-SEC/K9 2 x EHWIC-VA-DSL-B
The third ADSL line is connected over an ADSL modem at one fixed Router Gigabit interface.
I am trying to understand what load balancing method is used on a port channel on a Nexus switch . I have a server connected by a VPC to two Nexus switches. The nexus switches are only acting as layer 2 switches. I have a 6509 connected via a upstream link that does all of the routing for my VLANS. If have a server connected to the Nexus switches and it talks to a server on my 6509 what load balancing happens on the Nexus going across VPC 27 which is a layer 2 trunk going up to my 6509. Is it done on layer 2 or layer 3 flows?
My Nexus shows the default load balancing configurations
Port Channel Load-Balancing Configuration:System: source-dest-ip Port Channel Load-Balancing Addresses Used Per-Protocol:Non-IP: source-dest-macIP: source-dest-ip source-dest-mac
Is it possible to use two different load balancing methods at each end of a port-channel between two switches?
We have a Cisco 6509 at one end of the port-channel and a Cisco blade switch 3020 at the other end. Right now, we are using "src-dst-ip" at both end of the port-channel. We would like to change this. That is, we would like the #3020 switch to use "src-dst-ip" while the 6509 switch should use the "src-dst-port".
Why we want to do this, the reason is that we have FWSMs on the 6509. I've read that by configuring "src-dst-port" on the 6509, one can get a better performance of traffic going through the FWSM. However, the issue is that the 3020 switch does not support "src-dst-port".
I have a customer who wants his new ASA-5520 to load balance out-going traffic between 2 ISPs, fairly normal request. Now here's the twist. He wants to separate traffic based upon the protocol used, http to one ISP, https to the other.
Will there be some pause in traffic on formed ether channel interfaces (4500E switch), when i will change the default ether channel load balancing method to src-dst-port (or any other non-default method)?
I am testing on lab equipment (2 Catalyst 3550 and 1 Catalyst 3560) HSRP version 1 and 2.I successfully created a load balancing between the two Catalyst 3550 on a couple of vlans (11 and 12) on ver 1
now, just adding the command "standby xx version 2" my hosts on the 2 vlans are completely unable to ping the virtual IP def. gw on debugging i checked that msgs are exchangedthe two cat 3550 are seeing each other on HSRP (active / standby roles)the real ip addresses are pingable rebooted the swiches (just as a last resort try)deleted arp chache on hostsremoved the auth on hsrp all of this no effect.
i also tried to modify the priority on the cat 3560 (before he was on both vlans in standby) to make it the active one and with the same config it worked flawlessly.
My only idea is that there is a bug on CATs 3550 (IOS: c3550-ipservicesk9-mz.122-55.SE4.bin) [code]
I've 3750X switch that isn't loading email. then I went to rommon mode and accidently for "format flash". after that I loaded 15.0 SE2 s/w on it using tftp server but it doesn't boot up with that image. flash had only .bin file after I loaded it from tftp server.
since it wasn't booting up, I did format flash again and thought to load image again from tftp server but now, it doesn't load image from tftp server.
I am installing a several new Cisco VM servers. The VM hosts are losing connectivity when we apply the etherchannel config in the core stack. VMware has stated that the etherchannel load balance needs to be src-dst-ip in order for the etherchannel to work.However, my current stack has 2 etherchannels configured to other switches through out the bldg in network closets. The current load balance configuration is src-mac.My question....when I make the change to src-dst-ip, will my network connectivity go down? I saw there was an older article on this that doesnt provide an answer just a work around. Here is the article. [URL].
What is the load balance method of 3750 port channel ( by source ip , or by source mac ) to diver traffic to paths? I have tried to use 10.242.104.101 and 10.242.104.102 as source ip, it will travel to the same link (G0/1) within one port channel (G0/1+G0/2). Howerver, if I later use 10.242.104.109, then this time it will traffic to G0/2 link. What's the concept behind.
I've a network with 28 computers and 2 servers. Each server have a double Gbit port configured in Load Balancing & Fail Over.Now, I want to buy two Cisco's Switch SG 200-26 and I would know the best way to connect them and if it's possible to interconnect them with more than one cable to share the trafic.
1. Is this following solution a good one (does the link between swhitches will work when computers will access to servers) ?