Tie resolution in application load level balancing
AssigneeAlcatel Lucent
InventorRanjan Sharma
Life of this patent
Estimating the term…
Abstract
In a network such as a cellular telephone communications network, a control server balancing processing loads among several resource servers resolves a tie between resource servers using the resource servers' prior, current and future load state data and call gap percentages in each of the prior, current and future load states. Such tie resolution also takes into account the trend of resource demands placed on the network; whether it is monotonically increasing, monotonically decreasing or indeterminate. The goal of such resolution is to maximize the throughput of the network while eliminating random selection of resource servers to service requests.
Claims20
An apparatus, comprising
- a control server that controls tasks that the control server assigns to a plurality of resource servers, wherein prior to assigning a task to a resource server, said control server:
- receives a first status message comprising parameters that includes a call gap percentage of a next overload state of a first resource server and a previous overload state of the first resource server;
- receives a second status message comprising parameters that includes a call gap percentage of a next overload state of a second resource server and a previous overload state of the second resource server;
- and
- wherein the control server selects which of the first and second resource servers to send a task for processing by comparing the call gap percentage of the next overload state of the first resource server and the call gap percentage of the next overload state of the second resource server, and comparing the call gap percentage of the previous overload state of the first resource server and the call gap percentage of the previous overload state of the second resource server;
- and
- the control server communicates the task to the selected resource server, wherein a call gap percentage is a percentage of call requests that are refused service.
An apparatus, comprising:
- a resource server that processes an assigned task, said resource server sending at least one status message that comprises parameters that are indicative of:
- the current normal state or current overload state of the resource server;
- call gap percentages of at least the resource server's current overload state, next higher overload state, and previous overload state where a call gap percentage is a percentage of call requests that is refused service;
- where the resource server is in the continuum of either its current state or overload state;
- and
- where the resource server is assigned a task based on of the resource server call gap percentage in the next higher overload state, call gap percentage in the previous overload state and the call gap percentage of the current overload state as compared to the call gap percentages of a next higher overload state, the call gap percentages of previous overload state and the call gap percentages of current overload state of at least one other resource server.
A method for a control server to resolve a tie between resource servers in a network of servers, said method configured to execute on said control server and comprising the steps of:
- assigning by the control server, tasks to resource servers;
- receiving, at the control server, a first status message comprising parameters that includes a call gap percentage of a next and a previous overload state of a first resource server;
- receiving, at the control server, a second status message comprising parameters that includes a call gap percentage of a next and previous overload state of a second resource server;
- selecting, by the control server, one of the first and second resource servers to send a task for processing by comparing the call gap percentage of the next overload state of the first resource server and the call gap percentage of the next overload state of the second resource server, and comparing the call gap percentage of the previous overload state of the first resource server and a call gap percentage of the previous overload state of the second resource server;
- and
- communicating, by the control server, the task to the selected resource server, wherein a call gap percentage is a percentage of call requests that are refused service.
Dependencies are inferred from the claim text.