Server load balancing

Eildor

I started a similar thread to this in the past but don't think I explained myself as well as I could have, which caused some confusion.

I have a server connected to two layer 3 switches via two 4 gigabit EtherChannels (thats a 4 gigabit EtherChannel per switch). I want to send 50% of traffic down one EtherChannel link to one switch, and 50% of traffic down the other EtherChannel link to the other switch.

There must be an easy way to achieve this on a 3750-X, surely? Why does any configuration need to be done on the switches anyway? All I want to do is alternate between physical links.

Thanks.

malcybood

Hi Eildor,

I've read most of the post and the options you have here are basically as follows:

1. Create an 8GB etherchannel on a single 3750X stack (no FHRP if routed access is required). You can stack up to nine 3750X in a single stack.

Or you could have two stacks running at layer 2 with layer terminating of dual non vss distribution switches to let you configure a FHRP.

2. Implement Cisco Nexus 5000 switches with Cisco Nexus 2000 for server access. This solution allows you to configure VPC / VPC domain between the 5500 switches and would let let you configure the server with NIC teaming (single IP address).

You would still use FHRP as Nexus does not combine the switches as per VSS. Layer 2 VPC is similar to VSS but Layer 3 would still require a FHRP.

You can also connect the server to the Nexus 2000 with host VPC which allows you to connect a server at 2Gbps or 20Gbps etherchannel using LACP. This would be my preferred option if budget was available.

Data Center Routed Access Design with Cisco Nexus 5500 Platform and 2000 Series Fabric Extenders* [Cisco Nexus 5000 Series Switches] - Cisco Systems

3. Use a server load balancer such as Cisco ACE/GSS or F5 LTM/GTM to define virtual IPs and server farms along with DNS failover between DCs )depending on the topology).

This would be where you had 2 (or more) servers that replicate between each other within the DC for load sharing and failover. With an SLB appliance you can configure a virtual IP i.e. 10.10.10.1 then the two real server IPs - Server A 10.50.1.10, Server B 10.51.1.10 which are in separate server subnets in the server farm configuration.

This type of thing can be done within a single data centre or over 2 data centres.

Depends how much time you have to research this stuff yourself in time for your project and if there is a project budget, but I would suggest using one of the mentioned techniques.

I've been designing data centre hosting networks for quite a while now and very rarely see anything other than the above being designed / implemented.

You may also combine option 2 and 3, just depends on the requirements.