International Business Machines Corporation

Philip R Lee

Andrew Wright

Edward Alan Addison

Peggy A DeVal

Stephen John Burghard

Mark L Todd

Michael D Brooks

Catherine Mary Moxey

Robert Conroy Jones

Roger G Lewis

James A Harrison

Mayur Raja

Generating a performance policy for a server. A plurality of system metrics of a first server is monitored, and a benchmark for the plurality of system metrics based on the monitored plurality of system metrics is stored. The benchmark indicates desirable system metrics of the first server. An operating policy for a second server is generated from the stored benchmark. The operating policy includes desirable system metrics of the second server based on the desirable system metrics of the first server.

Server health checking

A system and computer program product for developing software in which the software comprises a plurality of programs. A change to a program is received. A data structure checking procedure may then be invoked. The changed program is parsed for a reference to a data structure. Other instances of the data structure are located in other programs within the software. The referenced data structure is compared to the located other instances of the data structure. A predefined action (such as notifying a programmer or correcting the inconsistencies) is performed in response to any detected differences between the referenced data structure and the located other instances of the data structure. These steps are repeated for all data structures within the changed program.

Maintaining consistency amongst data structures that are referenced within different programs

A method for developing software in which the software comprises a plurality of programs. A change to a program is received. A data structure checking procedure may then be invoked. The changed program is parsed for a reference to a data structure. Other instances of the data structure are located in other programs within the software. The referenced data structure is compared to the located other instances of the data structure. A predefined action (such as notifying a programmer or correcting the inconsistencies) is performed in response to any detected differences between the referenced data structure and the located other instances of the data structure. These steps are repeated for all data structures within the changed program.

Performance data for a healthy system is captured and used to create benchmarks that represent behavior of the computing environment in a healthy state. Current behavior (e.g, run-time characteristics) of the computing environment are monitored in an on-going manner, and are compared to the benchmarks. When a benchmark is violated, performance data capture is dynamically and automatically enabled. Or, if performance data is already being captured, then performance data capture may be enabled for a larger subset of possible data. Optionally, performance data capture may be automatically disabled upon detecting that the current behavior of the computing environment has returned to a healthy state where benchmarks are no longer violated.

Automated enablement of performance data collection

A multithreaded computing environment for capturing diagnostic data includes initiating a trap delay component having a delay threshold in response to a monitor determining that the computing environment is in a predetermined invalid state. The multithreaded computing environment for capturing diagnostic data also includes recording, by a first thread, diagnostic data for the computing environment. The trap delay component delays the step of recording diagnostic data until the monitor determines that the predetermined invalid state has been repeated for at least one of a total count or a period corresponding to a predetermined indicator that it is a state other than a transient state of the computing environment. The transient state corresponds to a transition of the computing environment by one or more other threads from a first valid state to a second valid state.

Diagnostic data capture in a computing environment

Identifying conflicting declared invalid runtime references of overlaid data structures of a shared memory area as declared in a program listing. The identifying includes identifying a first data structure and a first parameter list in the program listing. A first routine call and a sequential flow of execution of the first routine call through the program listing in order to write data to the shared memory area are identified. A second data structure and a second parameter list in the program listing are identified. A second routine call and a second sequential flow of execution of the second routine call through the program listing in order to write data to the shared memory area are identified. It is determined whether the second routine call is attempting to overwrite data of the first routine call with the data of the second routine call in the shared memory area.

Growing community of inventors

Winchester, United Kingdom

Philip R Lee