Transparent header modification for reducing serving load based on current and projected usage

摘要

A method and system for dynamically altering the delivery of web content to end users based on server load. Responsive to receiving a client request for web content, utilization metrics are collected from a plurality of devices that deliver the web content to the end users. Individual load levels for the devices are determined respectively, based on the utilization metrics of the devices, a combined load level is determined for two or more of the devices having the same device type based on the individual-load levels, and a service level to provide to the client is determined based on the combined load level. The request header is modified to specify a rate to deliver the web content to the client based on the service level. The request is sent with the modified header to one of the devices to serve a variant of the web content to the client at the specified rate.

主权项

1. A method of delivering web content to a client, responsive to a request from the client, the method comprising:
collecting utilization metrics from a plurality of devices that deliver web content to end users; determining, by one or more processors, individual load levels for each device in the plurality of devices, respectively, based on respective utilization metrics of the devices by: applying a device policy to the respective utilization metrics of each device in the plurality of devices by assigning, based on respective values of the respective utilization metrics, a weight value to each respective utilization metric; determining the individual load levels for each device in the plurality of devices based on weight values and a threshold level specified in the device policy, wherein the threshold level in the device policy specifies one of a highest assigned weight value, a lowest assigned weight value, or an average of assigned weight values of the respective utilization metrics; and writing the individual load levels to a memory; determining, by the one or more processors, a combined load level of two or more of the devices, in the plurality of devices, of a same device type based on the individual load levels for the two or more respective of the devices in a subset of devices in the plurality of devices; determining, by the one or more processors, a service level to provide to the client based on the combined load level; responsive to receiving the request from the client, modifying, by a layer 7 load balancer, a header in the request to specify a rate to deliver the web content to the client based on a determined service level, by: reading the determined service level from the memory; generating, using a transparent content negotiation protocol, a modified header based on the determined service level, wherein the modified header specifies a variant of the web content to deliver to the client; and inserting the modified header into the request; and sending the request with the modified header to one of the devices, wherein the one of the devices is directed to deliver the web content to the client at a specified rate.