10 Scheduling algorithms for LVS

1. Same big pot scheduling (Round-robin scheduling RR)
RR-Pure Polling method, compare garbage. Assign each request sequentially to the real server.


2. Same big pot dispatch with weights (Weighted round-robin scheduling WRR)
WRR-With a weighted polling method. Each request is sequentially dispatched in the real server, but more jobs are assigned to the larger server.

3. Assign who does not work (Least-connection LC)
LC-Dispatched according to the minimum number of connections

4. Who does not work with the weight assigned to whom (Weighted least-connections WLC default)
WLC-with weights. The machine is well-configured with high weights.

5. Region-based minimum connection scheduling (locality-based least-connection scheduling LBLC)
LBLC-cache server cluster. Local-based minimum connection. Pass the request to a server with a small load.

6. Least-region-based connection scheduling with replication schedule (locality-based least-connection scheduling with Replication scheduling LBLCR)
LBLCR-a cluster of cache servers with replication schedules. A page is cached on Server A, accessed at a high number of times, and other cache servers are less loaded, monitoring access to the same page, and if you are accessing the same page, divide the request to another server.

7. Target Hash dispatch (Destination hash scheduling DH)
Two IP is bound in Realserver. LD judges the ISP of the comer and transfers it to the appropriate IP.

8. Source hash scheduling (source hash scheduling SH)
Source Address hash. Differentiated based on the source of the client address. (rarely used)

9. Shortest expected delay (shortest expected delay scheduling SED)
Based on the WLC algorithm. This has to be for example.
ABC three machines weigh 123 respectively, and the number of connections is 123 respectively. Then if a new request comes in using the WLC algorithm, it may be assigned to any of the ABC. Such an operation is performed after using the SED algorithm
A: (+)/1
B: (1+2)/2
C: (1+3)/3
According to the result of the operation, the connection is given to C.
the algorithm that is understood by the individual is (1+ current number of connections)/weight.

10. Minimum queue scheduling (never queue scheduling NQ)
No queues are required. If there is a realserver number of connections = 0 is directly allocated in the past, do not need to do sed operation.

This article from "Cock Silk life" blog, declined reprint!

10 Scheduling algorithms for LVS