Sun Microsystems, Inc.

Google Inc.

Sybase, Inc.

Vitria Technology, Inc.

Other

Robert B Hagmann

Xiao Meng Zhang

Mark Tsimelzon

Dwight F Hare

Aleksey Sanin

Peter Vanderbilt

Eric S Tune

Vrijendra Gokhale

David M Brownell

Ian Marshall

Marion D Skeen

Kenneth M Cavanaugh, Iii

Anthony W Menges

Graham Hamilton

Alan Michael Snyder

Michael L Powell

Prashant Gupta

Rohit Jnagal

James E Van Riper

Katrina Garnett

Among other disclosed subject matter, a method includes receiving metric data associated with an execution of each of a plurality of task instances. The plurality of task instances include task instances associated with a task and the metric data for each task instance relating to execution performance of the task instance. The method includes for each task instance determining a deviation of the metric data associated with the task instance relative to an overall deviation of the metric data for the plurality of task instances of the task during each of a plurality of intervals and combining deviation measurements for the task instance that exceed a threshold deviation to obtain a combined deviation value. Each deviation measurement corresponds to the deviation of the metric data for one of the plurality of intervals. The method includes ranking the combined deviation values associated with at least a subset of the task instances.

Identifying task instance outliers based on metric data in a large scale parallel processing system

Among other disclosed subject matter, a computer-implemented method includes receiving an indication that execution of an instance of a first task is degraded relative to a performance threshold associated with the first task. Performance data associated with the execution of the instance of the first task and performance data associated with execution of a plurality of additional tasks executed on the computing device are collected. For each of the plurality of additional tasks, the method includes determining a score for the respective additional task based on the performance data associated with the execution of the instance of the first task and performance data associated with the respective additional task. The method includes identifying one or more additional tasks as a potential cause of degraded performance based on the scores associated with each of the plurality of additional tasks. The method includes modifying an execution of a particular identified task.

Identifying task instances that interfere with processor performance

The present invention provides a method for obtaining predicable and repeatable output results in a continuous processing system. The method involves processing messages and primitives in accordance with the following rules: (1) Messages are processed in accordance with timestamps, where messages are divided up into 'time slices'; (2) message order within a data stream is preserved among messages with the same time stamp; (3) subject to rule #, for each time slice, a primitive is executed when either the messages within such time slice show up in the input stream for such primitive or the state of the window immediately preceding such primitive changes due to messages within such time slice; and (4) for each time slice, primitives that are dependent on one or more upstream primitives are not executed until such upstream primitives have finished executing messages in such time slice that are queued for processing. If such rules are insufficient to determine the order in which primitives are processed, a deterministic 'tie-breaking' rule is then applied.

Method for obtaining repeatable and predictable output results in a continuous processing system

The present invention provides a system and method for measuring latency in a continuous processing system. The invention involves generating 'tracer' messages that are specifically designed for measuring latency. A tracer message is configured to pass through and be outputted by each primitive in the tracer message's path regardless of the function of the primitive and without changing the state of the primitive. Tracer messages are not filtered by filter primitives or delayed by time-delay primitives. Tracer messages are detected and discarded within loops. These rules provide for predictable behavior of tracer messages, thereby making them reliable indicators of latency and liveness in the continuous processing system. A tracer message is inserted into the continuous processing system at a desired start point for measuring latency. At a desired end point for measuring latency, latency is measured by calculating the amount of time the tracer message took to travel from the start point to the end point.

System and method for measuring latency in a continuous processing system

A method, in a continuous computation system, for processing a set of registered queries over a cluster of servers includes creating an execution plan for processing a set of queries over a cluster of servers, where the continuous computation system creates the execution plan by analyzing the semantics and requirements of the queries to determine how to distribute processing across the cluster. Analysis of a query can include determining whether input messages for the query can be processed independent of each other, whether input messages for the query can be partitioned into groups that can be processed independent of each other, whether the query includes an aggregator function, and whether the query includes a subquery.

Method for distributing processing of queries over a cluster of servers in a continuous processing system

Methods, apparatus and system programs are provided for holistic monitoring and troubleshooting an application where the application functionally depends upon a plurality of components on a network system and at least one of the plurality of components being selected from a group consisting of a network component, a hardware component, and a software component. The method includes collecting data from the components and analyzing data collected from the components to discover one or more issues in the components. The analyzing step includes considering domain knowledge of the components and considering the interrelationships and correlations between components working within the application. The method function includes diagnosing the issues in the components to determine an action plan.

System, method and model for autonomic management of enterprise applications

A number of methods, apparatus, and data structures are disclosed for managing computer processes. In one aspect, a daemon process which manages server processes includes an active server table and a locator service. The locator service can look up and register server processes in the active server table. Furthermore, the locator service can start up server processes. In some embodiments, the locator service includes a look-up object and a server process registration object which perform the tasks of the locator service. In other embodiments, methods for managing server process such as starting and registering the server processes are taught. In one specific method, a daemon process performs a variety of steps in response to receiving a look-up call for a target object. These steps include obtaining a server identifier for the target object, determining the state of a server process, and returning addressing information corresponding to the server process under which the target object will activate. In related method aspects the daemon process will start the server process if it isn't running and/or wait until the server process is running to return the addressing information. In a separate method aspect, a server process self-starts; receiving an object reference for a desired target object, receiving a server process identification number, creating a communications port for itself, forming addressing information for itself, obtaining an object reference for a server process registration object, and registering itself by calling the server process registration object to invoke a register new process operation.

Methods and apparatus for managing computer processes

A computer-implemented process for accelerating operations of a real time decision support system executing on a computer including the steps of determining the elements of a query, determining sources of data which are reference collections for a query, determining if a query contains subexpressions over reference collections which are sources of data for the query, precomputing results for functions on data in reference collections which are sources of data for the query, and caching the precomputed results.

Precomputing reference collections in a decision support system

A variety of methods, apparatus and data structures for managing collections of objects are described. In one aspect of the invention, an object that is intended for use in a distributed object operating environment has a structure including a group designation, a co-activation designation and a co-process designation. The group designation is arranged to identify a group to which the object belongs. The group is defined as a collection of objects which share a common persistent state. The co-activation designation is arranged to identify a co-activation set to which the object belongs. The co-activation set is a collection of objects which are to be activated at the same time. The co-process designation is arranged to identify a co-process set to which the object belongs. The co-process set is a collection of objects which are to be activated within a single process. A various embodiments, a variety of methods of utilizing one or more of these designations to facilitate efficient operation of a distributed computing system are also described. In some applications, a particular object may be conceptually divided into a plurality of sub-objects, with each sub-object having its own portion of persistent memory. In this embodiment, the particular object may only be invoked as a whole, but the object is provided with a mechanism for accessing the selected sub-object in response to a call from a client object that invokes the object and identifies the sub-object in a sub-object field of an object reference that refers to the object. When sub-objects are use, the object references may be arranged to include a host identifier, an object identifier and a sub-object field.

Methods and apparatus for managing collections of objects

A computer-implemented process for evaluating events furnished to a real time decision support system executing on a computer with respect to queries submitted by users comprising the steps of parsing each query into its elements to determine its correctness, eliminating incorrect queries, performing single query optimizations on each query, performing multiquery optimizations on all queries for each event type, and performing result processing optimizations for each query.

Real-time query optimization in a decision support system

An apparatus and a method are disclosed whereby a client application can use a trusted 'Deputy' application to execute operation calls on several servers on behalf of the client, the Deputy application being able to convince the servers that the deputy application is representing the original client application, and the client thereby reducing the risk of being contaminated by uncontrolled access to an unknown server, while at the same time being able to obtain the desired processing results regardless of the number or location of servers involved in providing the results. The Deputy application is authenticated by the server as representing a user, not a user on a predetermined workstation or as a predetermined member of a particular work group.

Growing community of inventors

Palo Alto, CA, United States of America

Robert B Hagmann