Web Api建议重定向呼叫，负载平衡

Question

I have my Web Api application distributed on three servers using Load Balancer with Round Robin rule. Every server can proceed only one request for specified RestApi endpoint http://MyServer/api/LongRunningProcees If second request is posted to the same Rest API endpoint when first one is not finished a server return 503 Service unavailable.

I am new to load balancing and I don't find any solution how to redirect the request to another server until it will be accepted. Is this Load Balancer responsibility?

Answer 1

If you want to use Round Robin algorithm then you have to accept that some servers will be busy while others will be waiting for requests. This algorithm simply works in this way.

However, there are also other algorithms. For example Least response time which selects servers based on the response time from the servers. The details depends on the used Load Balancer.

Besides, it seems to me that a server should not return 503 error if it is busy. Instead a request should be put into some kind of queue and wait until a server is available. If requests are processed in a similar time and an suitable balancing algorithm is used then all server should have the similar number of requests in their queues.