by Aidan Clarke
Traffic IP (TIP) Groups are the key to how Brocade vADC delivers intelligent clustering, as we saw in the previous article. This is because we handle Traffic IP Addresses in a novel way - Traffic IP (TIP) Groups allow you to define one or more IP addresses to be used for an application endpoint, and you can decide how you want these IP addresses distributed across your cluster.
This means that you are not restricted to a single IP address allocated to a Virtual Server. Once your virtual server is configured, you simply bind it to one or more Traffic IP Groups, which gives you much greater flexibility in how you design your service.
- A Traffic IP (TIP) Group is a “Listener” for incoming traffic to be “Load Balanced”
- TIP Groups can have one or more Traffic IP addresses
- TIP Groups can live on one or multiple Traffic Managers
- By default, TIP Groups are “Round Robin” distributed across the cluster
|
Service Example: |
With Brocade vADC: |
|
Application #1 must be Active/Active/Active on three nodes of a three-node cluster, but you want a warm standby node? |
Create a TIP Group with three IP addresses on a four-node vTM cluster with one of the nodes marked as passive (T1 above) |
|
Application #2 needs a simple Active/Standby configuration with a single IP address? |
Create a simple TIP group with one of the nodes as passive (T2 above) |
|
Application #3 must be Active/Active/Standby on a three-node cluster? |
Create a TIP Group with two IP addresses on a three-node cluster and (optionally) mark one of the cluster nodes as passive (T3 above) |
|
Application #4 must be active on all nodes of an eight node cluster but only have one IP address? |
Use a multicast enabled environment and deploy a multi-hosted Traffic IP Group. vADC will ensure the workload is evenly distributed across the cluster |
|
All of the above, at the same time for different Virtual Servers on the same cluster? |
No problem. Brocade vADC excels at all aspects of clustering |
What happens when a node fails?
Brocade vADC monitors the health on all nodes, and when a node fails, then each service is reconfigured gracefully. In this example, if the second node fails, then:
- Application #1 transfers workload to standby (node #4)
- Application #2 transfers workload to standby (node #3)
- Application #3 continues unchanged, but shares workload on node #4
During failover, session mapping is maintained for all nodes that remain active; and as standby nodes are brought online, session persistence can pin new user sessions to the new nodes. For recovery, or for planned maintenance, a node can be “drained” - no new sessions are pinned to the node - allowing for graceful reconfiguration or shutdown.
Traffic IP Groups allow for flexible allocation of workload as applications are brought online or reconfigured across a cluster, allowing you to express the relative priorities of applications in a shared infrastructure, an the desired behavior for failover and recovery.
And by increasing the number of cluster nodes dynamically, the scope of a node failure is reduced. By reducing the size of fault domains, services can recover more quickly, and additional nodes can be brought online to return services to full resilience.
