In support of load balancing and high-availability functions, the Universal Compute Platform relies on Virtual Router Redundancy Protocol (VRRP) to provide IP abstraction to key functionality. VRRP is critical in the configuration model.
The node with the higher priority defaults to the master. However, in the case of failovers of the master node, VRRP algorithms assign mastery to the next higher priority member of the cluster. Therefore, it is important to properly assign corresponding priority settings to each node, so that their hierarchical priority in terms of VRRP state ownership is clear.
As a best practice:
An Inter-Cluster Connection refers to the back-end interaction and synchronization between all the members of a cluster. Minimum required connection requires 10 Gbps between nodes.
| Nodes 1 -6 (Port #) | |||
|---|---|---|---|
| ICC |
|
||
| VLAN | VLAN Tagged/Untagged | ||
| Port type | Physical | ||
| VRRP | |||
| VRRP IP addresses | ICC VRRP IP | ||
| Priority | Set a unique priority for each node.
For example:
|
||
| Router ID | ID (2) | ||
The VRRP configuration relates to the number of services you are exposing. Configure a VRRP IP address (VIP) for each service.
| Nodes 1-6 (Port #) | |
|---|---|
| Data Port | Node Port IP /CIDR. Unique Port IP for each node. |
| VLAN | VLAN Tagged/Untagged |
| Port type | Physical |
| VRRP | |
| VRRP IP address (VIP) | 6 VIP addresses. Unique VIP for each node |
| Priority | Unique priority value for each VIP |
| Router ID | ID (1) |