I would like to know how the provided policy options in the Clustering Balancer work and what are their differences:
Round Robin
Weighted Round Robin
Least Connections
Weighted Least Connections
Locality Based Least Connections
I already looked at the Cluster Docs and this forum looking for answers and none were found.
Where could I find this information?
Thanks and Regards,
Rodrigo
You can find it right here. In this very post!
Round Robin
The round robin algorithm maintains a list of servers and forwards each new connection to the next server in the list.
Weighted Round Robin
This load-balancing algorithm works like round robin except you can favor certain nodes in the array, by setting a proportional weight. The higher the weight number, the more connections will be forwarded to that server. Useful for tasking a server with a slightly faster CPU to handle a few more requests if your nodes are asymmetric.
Least Connections
Least connections does what it says on the box - InterWorx knows which nodes have more connections than others, and it factors this into where to send connections, dropping them on the server with the fewest connections open right now.
Weighted Least Connections
In weighted least-connections load balancing, an incoming network connection is assigned to the server which has the smallest current active connection number to weight ratio.
Locality Based Least Connections
This one is where it gets … interesting. Locality-based least connections load balancing usually directs a connection destined for an IP address to its server if the server is alive and under load. If the server is overloaded (its active connection number is larger than its weight) and there is one or more other server in a partial-load state, then the algorithm will allocate the weighted least-connection server to the incoming client IP. The goal of this algorithm is that when load on all the servers are almost balanced, sending requests for the same client IP address to the same server will fully take advantage of access locality and improve cache hit rate at the server, improving the throughput of the whole cluster.