Load balancer host selection and fault detection

A computer-implemented method for distributing processing load, said method comprising:

• under the control of one or more computer systems configured with executable instructions,
• distributing a set of requests from a load balancer to a plurality of services for processing the requests;
• monitoring a state of the requests being processed by the plurality of services by a load balancer over a specified time interval;
• determining that at least one service in the plurality of services has deviated from remaining services in the plurality of services with respect to the state being monitored by the load balancer over the specified time interval, wherein determining that the at least one service in the plurality of services has deviated further includes determining that a sampled response time for the at least one service is shorter in duration than sampled response times for other services;
• and
• suspending the distribution of the requests from the load balancer to the service that has deviated from the remaining services.

A computing device, comprising:

• a processor;
• and
• memory including instructions that, when executed by the processor, cause the computing device to:
• receive a set of requests by a load balancer and route each of the set of requests to at least one of a plurality of services for processing;
• record, by the load balancer, a state of the requests being processed by the plurality of services over a specified time interval;
• determine that at least one service in the plurality of services has deviated from remaining services in the plurality of services with respect to the state being monitored by more than a predetermined threshold amount, wherein to determine that the at least one service in the plurality of services has deviated further includes determining that a sampled response time for the at least one service is shorter in duration than sampled response times for the remaining services;
• and
• flag the service that has deviated from the remaining services as behaving abnormally.

A non-transitory computer readable storage medium storing a set of instructions, the instructions when executed by one or more hardware processors, causing the one or more hardware processors to perform a set of steps comprising:

• distributing a set of requests from a load balancer to a plurality of services for processing the requests;
• maintaining a state of the requests being processed by the plurality of services by a load balancer over a specified time interval;
• detecting that at least one service in the plurality of services has deviated, by more than a predetermined threshold, from remaining services in the plurality of services with respect to the state being monitored by the load balancer, wherein detecting that the at least one service in the plurality of services has deviated further includes determining that a sampled response time for the at least one service is shorter in duration than sampled response times for the remaining services;
• and
• disabling the distribution of the requests from the load balancer to the service that has deviated from the remaining services.