Autonomic assignment of communication buffers by aggregating system profiles
View Patent ↗A method, system and apparatus for autonomic buffer configuration. In accordance with the present invention, an autonomic buffer configuration method can include monitoring data flowing through buffers in a communications system and recording in at least one buffer profile different data sizes for different ones of the data flowing through the buffers during an established interval of time. An optimal buffer size can be computed based upon a specification of a required percentage of times a buffer must be able to accommodate data of a particular size. Subsequently, at least one of the buffers can be re-sized without re-initializing the at least one resized buffer.
1. A profile processor device disposed within an autonomic buffer configuration system comprising:
a performance monitor arranged to monitor data flowing through at least one buffer in a communications system;
a reporting tool configured to generate at least one buffer profile based upon monitored information produced by said performance monitor;
a buffer size calculator programmed to compute an optimal buffer size for said at least one buffer based upon said at least one buffer profile; and,
a buffer resizing component coupled to said at least one buffer and programmed to dynamically resize said at least one buffer to said optimal buffer size without re-initializing said buffer.
2. The profile processor device of claim 1 , wherein said at least one buffer is selected from the group consisting of a profile processor application-level buffer and a kernel-level buffer.
3. The profile processor device of claim 1 , wherein said data comprises at least one of requests and responses to said requests.
4. The profile processor device of claim 1 , wherein said communications system is disposed within one of a Web server and an applications server.
5. The profile processor device of claim 2 , wherein said performance monitor comprises a configuration for performing an analysis of an amount of data passed between said application-layer buffer and said kernel-layer buffer.
6. The profile processor device of claim 3 , wherein said performance monitor comprises a configuration for performing at least one of (1) a statistical analysis of request sizes for an interval of time for said communications system, (2) a statistical analysis of request sizes for an interval of time for individual connections in said communications system, (3) a statistical analysis of inserting delay durations of varying lengths between consecutive input/output operations in said communications system, and (4) a statistical analysis of patterns of requests and an ordering of said requests in said patterns.
7. The profile processor device of claim 1 , wherein said monitored information is weighted in said at least one buffer profile.
8. The profile processor device of claim 1 , further comprising a profile aggregator configured to combine individual buffer profiles to produce a single profile for use by said calculator in computing an optimal buffer size.