Query task processing based on memory allocation and performance criteria

Abstract

Described are methods, systems and computer readable media for query task operations based on memory allocation and performance criteria.

Description

Embodiments relate generally to computer data systems, and more particularly, to methods, systems and computer readable media for memory-efficient and processor-efficient aggregation and data source refreshing.

Data sources or objects within a computer data system may include static sources and dynamic sources. Some data sources or objects (e.g., tables) may depend on other data sources. As new data is received or obtained for dynamic data sources, those dynamic data sources may be refreshed (or updated). Data sources or objects that are dependent on one or more dynamic sources that have been refreshed may also need to be refreshed. The refreshing of data sources may need to be performed in an order based on dependencies.

In addition, some data sources or objects (e.g., tables) contain logical arrays (or collections) of values from other data sources. Data from the other data sources used to create the logical arrays may change. The logical arrays may need to be refreshed as the data used to create the arrays are changed.

Embodiments were conceived in light of the above mentioned needs, problems and/or limitations, among other things.

Some implementations (first implementations) can include a system for maximizing memory and processing efficiencies in a computer system, the system comprising one or more processors and computer readable storage coupled to the one or more processors, the computer readable storage having stored thereon instructions that, when executed by the one or more processors, cause the one or more processors to perform operations. The operations can include sending a digital request from a client computer to a remote query processor on a query server computer. The operations can include creating and storing, in a computer storage, data in a plurality of column sources. The operations can also include creating and storing in a computer memory a first table object index mapping data in the plurality of column sources to a first table object, the computer memory having faster access time than the computer storage. At the remote query processor, memory-efficient and processor-efficient operations can be provided including an aggregation query operation. The aggregation query operation can include receiving an aggregation query task to create a second table object comprising an updateable aggregation of rows from the first table object. The aggregation query operation can include creating and storing in the computer memory separate from the computer storage, the second table object. The aggregation query operation can also include creating a plurality of aggregation states for a plurality of aggregation rows of the second table object, each of the plurality of aggregation rows representing an aggregation of one or more rows of the first table object, each of the plurality of aggregation states being configured to determine an aggregated value for a respective one of the plurality of aggregation rows by executing an aggregation formula. The aggregation query operation can further include storing determined aggregated values in the computer memory separate from the computer storage. The aggregation query operation can include creating and starting a listener for the second table object. The aggregation query operation can also include subscribing to receive at the listener an indication of an update to the first table object. The aggregation query operation can include, responsive to receiving the indication at the listener, selecting one or more of the aggregation states to update the second table object and accessing at most only a subset of rows of the first table object, thereby eliminating the use of repeat processing time for re-executing the aggregation formula and providing faster access to the plurality of aggregation rows of the second table object. The computer storage can be remote from the remote query processor and can be accessed by the remote query processor via a network. The computer memory can be local to the remote query processor.

In some first implementations, aggregated values comprise a collection for each aggregated column of each row of the second table object. In some first implementations, a first collection of a first aggregated column of a first row of the second table object is a first ordered collection, a second collection of a second aggregated column of the first row of the second table object is a second ordered collection, and the first ordered collection and the second ordered collection have a same ordering. In some first implementations, the collection comprises a reference to data of the first table object. In some first implementations, the aggregation function determines a summation and the aggregated values comprise a running sum. In some first implementations, the aggregation function determines a standard deviation and the aggregated values comprise a running sum, a running sum of squares, and a count.

In some first implementations, the aggregation query operation further includes, responsive to the receiving at the listener, detecting a change to an output of an aggregation state of the plurality of aggregation states; and transmitting, responsive to the detecting, a change notification to a subscriber of the second table object. In some first implementations, the change notifications can include one of a data add notification, a data modify notification, a data delete notification and a data reindex notification. In some first implementations, at least one of the plurality of aggregation states is configured to determine an aggregated value based on a new value and an associated previous value of the first table object. In some first implementations, the first ordering and the second ordering is determined based on an ordering of the first table object. In some first implementations, the aggregation query operation further includes determining a mapping of a subset of columns and applying the mapping to one or more other columns to create one or more of the collections for the aggregated columns of the second table object.

Some implementations (second implementations) can include a system for maximizing memory and processing efficiencies in a computer system, the system comprising one or more processors and computer readable storage coupled to the one or more processors, the computer readable storage having stored thereon instructions that, when executed by the one or more processors, cause the one or more processors to perform operations. The operations can include storing a second table object representing an updateable aggregation of rows from a first table object. The operations can include creating a plurality of aggregation states for a plurality of aggregation rows of the second table object, each of the plurality of aggregation rows representing an aggregation of one or more rows of the first table object, each of the plurality of aggregation states being configured to determine an aggregated value for a respective one of the plurality of aggregation rows by executing an aggregation formula. The operations can also include storing determined aggregated values in the computer memory separate from the computer storage. The operations can include receiving an indication of an update to the first table object. The operations can further include, responsive to receiving, selecting one or more of the aggregation states to update the second table object and accessing at most only a subset of rows of the first table object, thereby eliminating the use of repeat processing time for re-executing the aggregation formula and providing faster access to the plurality of aggregation rows of the second table object.

In some second implementations, the aggregated values comprise a collection for each aggregated column of each row of the second table object. In some second implementations, a first collection of a first aggregated column of a first row of the second table object is a first ordered collection, a second collection of a second aggregated column of the first row of the second table object is a second ordered collection, and the first ordered collection and the second ordered collection have a same ordering. In some second implementations, the collection comprises a reference to data of the first table object. In some second implementations, the aggregation function determines a weighted average. In some second implementations, a combination aggregation is determined for a row of the second table object by two or more of the plurality of aggregation states. In some second implementations, data of the first table object is stored in a computer storage that is remote from the computer memory and is accessed via a network, and the computer memory is local to the one or more processors. In some second implementations, the operations can further include determining a mapping of a subset of columns and applying the mapping to one or more other columns to create one or more of the collections for the aggregated columns of the second table object.

In some second implementations, the operations further include detecting a change to an output of an aggregation state of the plurality of aggregation states; and transmitting, responsive to the detecting, a change notification to a subscriber of the second table object. In some second implementations, the change notifications can include one of a data add notification, a data modify notification, a data delete notification and a data reindex notification. In some second implementations, at least one of the plurality of aggregation states is configured to determine an aggregated value based on a new value and an associated previous value of the first table object. In some second implementations, the first ordering and the second ordering is determined based on an ordering of the first table object.

Some implementations (third implementations) can include a nontransitory computer readable medium having stored thereon software instructions that, when executed by one or more processors, cause the one or more processors to perform operations. The operations can include storing, in a computer memory, a second table object representing an updateable aggregation of rows from a first table object. The operations can include creating, at a processor coupled to the computer memory, a plurality of aggregation states for a plurality of aggregation rows of the second table object, each of the plurality of aggregation rows representing an aggregation of one or more rows of the first table object, each of the plurality of aggregation states being configured to determine an aggregated value for a respective one of the plurality of aggregation rows by executing an aggregation formula. The operations can also include storing determined aggregated values in the computer memory. The operations can further include receiving an indication of an update to the first table object. The operations can also include, responsive to receiving, selecting one or more of the aggregation states to update the second table object and accessing a small subset of rows of the first table object, thereby eliminating the use of repeat processing time for re-executing the aggregation formula and providing faster access to the plurality of aggregation rows of the second table object.

In some third implementations, the aggregated values comprise a collection for each aggregated column of each row of the second table object. In some third implementations, a first collection of a first aggregated column of a first row of the second table object is a first ordered collection, a second collection of a second aggregated column of the first row of the second table object is a second ordered collection, and the first ordered collection and the second ordered collection have a same ordering. In some third implementations, the collection comprises a reference to data of the first table object. In some third implementations, the aggregation function determines a maximum value. In some third implementations, data of the first table object is stored in a computer storage that is remote from the computer memory and is accessed via a network, and the computer memory is local to the one or more processors. In some third implementations, the operations further include determining a mapping of a subset of columns and applying the mapping to one or more other columns to create one or more of the collections for the aggregated columns of the second table object.

In some third implementations, the operations further include detecting a change to an output of an aggregation state of the plurality of aggregation states; and transmitting, responsive to the detecting, a change notification to a subscriber of the second table object. In some third implementations, the change notifications can include one of a data add notification, a data modify notification, a data delete notification and a data reindex notification. In some third implementations, at least one of the plurality of aggregation states is configured to determine an aggregated value based on a new value and an associated previous value of the first table object. In some third implementations, the first ordering and the second ordering is determined based on an ordering of the first table object.

Some implementations (fourth implementations) can include a method for maximizing memory and processing efficiencies in a computer system. The method can include storing, in a computer memory, a second table object representing an updateable aggregation of rows from a first table object. The method can also include creating, at a processor coupled to the computer memory, a plurality of aggregation states for a plurality of aggregation rows of the second table object, each of the plurality of aggregation rows representing an aggregation of one or more rows of the first table object, each of the plurality of aggregation states being configured to determine an aggregated value for a respective one of the plurality of aggregation rows. The method can also include storing determined aggregated values in the computer memory. The method can also include receiving an indication of an update to the first table object. The method can also include, responsive to receiving, selecting one or more of the aggregation states to update the second table object and accessing at most only a subset of rows of the first table object, thereby eliminating the use of repeat processing time for re-executing the aggregation formula and providing faster access to the plurality of aggregation rows of the second table object.

In some fourth implementations, the aggregated values comprise a collection for each of one or more aggregated columns of each row of the second table object. In some fourth implementations, a first collection of a first aggregated column of a first row of the second table object is a first ordered collection, a second collection of a second aggregated column of the first row of the second table object is a second ordered collection, and the first ordered collection and the second ordered collection have a same ordering. In some fourth implementations, the collection comprises a reference to data of the first table object. In some fourth implementations, the method further includes detecting a change to an output of an aggregation state of the plurality of aggregation states; and transmitting, responsive to the detecting, a change notification to a subscriber of the second table object. In some fourth implementations, the change notifications can include one of a data add notification, a data modify notification, a data delete notification and a data reindex notification. In some fourth implementations, at least one of the plurality of aggregation states is configured to determine an aggregated value based on a new value and an associated previous value of the first table object. In some fourth implementations, the first ordering and the second ordering is determined based on an ordering of the first table object. In some fourth implementations, the method further includes determining a mapping of a subset of columns and applying the mapping to one or more other columns to create one or more of the collections for the aggregated columns of the second table object. In some fourth implementations, data of the first table object is stored in a computer storage that is remote from the computer memory and is accessed via a network, and the computer memory is local to the one or more processors.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a diagram of an example computer data system showing an example data distribution configuration in accordance with some implementations.

FIG. 2 is a diagram of an example computer data system showing an example administration/process control arrangement in accordance with some implementations.

FIG. 3 is a diagram of an example computing device configured for memory-efficient and processor-efficient aggregation and data source refresh processing in accordance with some implementations.

FIG. 4 is a flowchart of an aggregation operation in accordance with some implementations.

FIGS. 5A-H illustrate source and result tables of example aggregation operations in accordance with some implementations.

FIGS. 6A-C illustrate an example performing multiple aggregations together in accordance with some implementations.

FIGS. 7A and 7B show data source definitions and a corresponding directed acyclic graph (DAG) in accordance with some implementations.

FIG. 8A shows a data source table (t1 defined in FIG. 7A) in accordance with some implementations.

FIGS. 8B and 8C illustrate an aggregation table that is a result of the aggregation operation shown in FIGS. 7A-7B in accordance with some implementations.

FIG. 9A illustrates automatic updating of an aggregation result when a row is modified in a source table in accordance with some implementations.

FIG. 9B illustrates automatic updating of an aggregation result when a row is modified in a source table and the modification includes a change to the key in accordance with some implementations.

FIG. 9C illustrates automatic updating of an aggregation result when a row is modified in a source table and the modification includes a change to the key in accordance with some implementations.

FIG. 9D illustrates automatic updating of an aggregation result when a row is added to a source table in accordance with some implementations. When row 906 is added to t1 is added as shown in FIG. 9D, t2 is notified via an AMDR message.

FIG. 9E illustrates automatic updating of an aggregation result when a row is deleted from a source table in accordance with some implementations.

FIG. 10 is a diagram of an example computer data system showing an example data distribution configuration in accordance with some implementations.

FIG. 11 is a diagram of an example computer data system showing an example data distribution configuration in accordance with some implementations.

FIG. 12 is a flowchart of an example aggregation operation in accordance with some implementations.

FIG. 13 is a diagram of an example computer data system showing an example data distribution configuration in accordance with some implementations.

FIGS. 14A and 14B also show data source definitions and a corresponding directed acyclic graph (DAG) in accordance with some implementations.

DETAILED DESCRIPTION

Reference may be made herein to the Java programming language, Java classes, Java bytecode and the Java Virtual Machine (JVM) for purposes of illustrating example implementations. It will be appreciated that implementations can include other programming languages (e.g., groovy, Scala, R, Go, etc.), other programming language structures as an alternative to or in addition to Java classes (e.g., other language classes, objects, data structures, program units, code portions, script portions, etc.), other types of bytecode, object code and/or executable code, and/or other virtual machines or hardware implemented machines configured to execute a data system query.

FIG. 1 is a diagram of an example computer data system and network 100 showing an example data distribution configuration in accordance with some implementations. In particular, the system 100 includes an application host 102, a periodic data import host 104, a query server host 106, a long-term file server 108, and a user data import host 110. While tables are used as an example data object in the description below, it will be appreciated that the data system described herein can also process other data objects such as mathematical objects (e.g., a singular value decomposition of values in a given range of one or more rows and columns of a table), TableMap objects, etc. A TableMap object provides the ability to lookup a Table by some key. This key represents a unique value (or unique tuple of values) from the columns aggregated on in a byExternal( ) statement execution, for example. A TableMap object can be the result of a byExternal( ) statement executed as part of a query. It will also be appreciated that the configurations shown in FIGS. 1 and 2 are for illustration purposes and in a given implementation each data pool (or data store) may be directly attached or may be managed by a file server.

The application host 102 can include one or more application processes 112, one or more log files 114 (e.g., sequential, row-oriented log files), one or more data log tailers 116 and a multicast key-value publisher 118. The periodic data import host 104 can include a local table data server, direct or remote connection to a periodic table data store 122 (e.g., a column-oriented table data store) and a data import server 120. The query server host 106 can include a multicast key-value subscriber 126, a performance table logger 128, local table data store 130 and one or more remote query processors (132, 134) each accessing one or more respective tables (136, 138). The long-term file server 108 can include a long-term data store 140. The user data import host 110 can include a remote user table server 142 and a user table data store 144. Row-oriented log files and column-oriented table data stores are discussed herein for illustration purposes and are not intended to be limiting. It will be appreciated that log files and/or data stores may be configured in other ways. In general, any data stores discussed herein could be configured in a manner suitable for a contemplated implementation.

In operation, the input data application process 112 can be configured to receive input data from a source (e.g., a securities trading data source), apply schema-specified, generated code to format the logged data as it's being prepared for output to the log file 114 and store the received data in the sequential, row-oriented log file 114 via an optional data logging process. In some implementations, the data logging process can include a daemon, or background process task, that is configured to log raw input data received from the application process 112 to the sequential, row-oriented log files on disk and/or a shared memory queue (e.g., for sending data to the multicast publisher 118). Logging raw input data to log files can additionally serve to provide a backup copy of data that can be used in the event that downstream processing of the input data is halted or interrupted or otherwise becomes unreliable.

A data log tailer 116 can be configured to access the sequential, row-oriented log file(s) 114 to retrieve input data logged by the data logging process. In some implementations, the data log tailer 116 can be configured to perform strict byte reading and transmission (e.g., to the data import server 120). The data import server 120 can be configured to store the input data into one or more corresponding data stores such as the periodic table data store 122 in a column-oriented configuration. The periodic table data store 122 can be used to store data that is being received within a time period (e.g., a minute, an hour, a day, etc.) and which may be later processed and stored in a data store of the long-term file server 108. For example, the periodic table data store 122 can include a plurality of data servers configured to store periodic securities trading data according to one or more characteristics of the data (e.g., a data value such as security symbol, the data source such as a given trading exchange, etc.).

The data import server 120 can be configured to receive and store data into the periodic table data store 122 in such a way as to provide a consistent data presentation to other parts of the system. Providing/ensuring consistent data in this context can include, for example, recording logged data to a disk or memory, ensuring rows presented externally are available for consistent reading (e.g., to help ensure that if the system has part of a record, the system has all of the record without any errors), and preserving the order of records from a given data source. If data is presented to clients, such as a remote query processor (132, 134), then the data may be persisted in some fashion (e.g., written to disk).

The local table data server 124 can be configured to retrieve data stored in the periodic table data store 122 and provide the retrieved data to one or more remote query processors (132, 134) via an optional proxy.

The remote user table server (RUTS) 142 can include a centralized consistent data writer, as well as a data server that provides processors with consistent access to the data that it is responsible for managing. For example, users can provide input to the system by writing table data that is then consumed by query processors.

The remote query processors (132, 134) can use data from the data import server 120, local table data server 124 and/or from the long-term file server 108 to perform queries. The remote query processors (132, 134) can also receive data from the multicast key-value subscriber 126, which receives data from the multicast key-value publisher 118 in the application host 102. The performance table logger 128 can log performance information about each remote query processor and its respective queries into a local table data store 130. Further, the remote query processors can also read data from the RUTS, from local table data written by the performance logger, or from user table data read over NFS, for example.

It will be appreciated that the configuration shown in FIG. 1 is a typical example configuration that may be somewhat idealized for illustration purposes. An actual configuration may include one or more of each server and/or host type. The hosts/servers shown in FIG. 1 (e.g., 102-110, 120, 124 and 142) may each be separate or two or more servers may be combined into one or more combined server systems. Data stores can include local/remote, shared/isolated and/or redundant. Any table data may flow through optional proxies indicated by an asterisk on certain connections to the remote query processors. Also, it will be appreciated that the term “periodic” is being used for illustration purposes and can include, but is not limited to, data that has been received within a given time period (e.g., millisecond, second, minute, hour, day, week, month, year, etc.) and which has not yet been stored to a long-term data store (e.g., 140).

FIG. 2 is a diagram of an example computer data system 200 showing an example administration/process control arrangement in accordance with some implementations. The system 200 includes a production client host 202, a controller host 204, a GUI host or workstation 206, and query server hosts 208 and 210. It will be appreciated that there may be one or more of each of 202-210 in a given implementation.

The production client host 202 can include a batch query application 212 (e.g., a query that is executed from a command line interface or the like) and a real time query data consumer process 214 (e.g., an application that connects to and listens to tables created from the execution of a separate query). The batch query application 212 and the real time query data consumer 214 can connect to a remote query dispatcher 222 and one or more remote query processors (224, 226) within the query server host 1208.

The controller host 204 can include a persistent query controller 216 configured to connect to a remote query dispatcher 232 and one or more remote query processors 228-230. In some implementations, the persistent query controller 216 can serve as the “primary client” for persistent queries and can request remote query processors from dispatchers, and send instructions to start persistent queries. For example, a user can submit a query to 216, and 216 starts and runs the query every day. In another example, a securities trading strategy could be a persistent query. The persistent query controller can start the trading strategy query every morning before the market opened, for instance. It will be appreciated that 216 can work on times other than days. In some implementations, the controller may require its own clients to request that queries be started, stopped, etc. This can be done manually, or by scheduled (e.g., cron) jobs. Some implementations can include “advanced scheduling” (e.g., auto-start/stop/restart, time-based repeat, etc.) within the controller.

The GUI/host workstation can include a user console 218 and a user query application 220. The user console 218 can be configured to connect to the persistent query controller 216. The user query application 220 can be configured to connect to one or more remote query dispatchers (e.g., 232) and one or more remote query processors (228, 230).

FIG. 3 is a diagram of an example computing device 300 in accordance with at least one implementation. The computing device 300 includes one or more processors 302, operating system 304, computer readable medium 306 and network interface 308. The memory 306 can include a data source refresh application 310 and a data section 312 (e.g., for storing DAGs, etc.).

In operation, the processor 302 may execute the application 310 stored in the memory 306. The application 310 can include software instructions that, when executed by the processor, cause the processor to perform operations for memory-efficient and processor-efficient aggregation and data source refreshing in accordance with the present disclosure (e.g., performing one or more of 402-410 and 1202-1226 described below).

The application program 310 can operate in conjunction with the data section 312 and the operating system 304.

FIG. 4 is a flowchart of an aggregation operation in accordance with some implementations. Processing begins at 402.

At 402, an aggregation operation is called on a source table. The aggregation operation can specify zero or more “groupColumns” of the source table. The aggregation operation can, for example, be a general “by( )” operation, a dedicated aggregator, a combined aggregation, a “byExternal( )” operation, a user-defined aggregator, or the like. In some implementations, the syntax for calling a “by( )” aggregation operation is “Table by(String . . . groupColumns)”. Processing continues to 404.

At 404, the columns defined by groupColumns (i.e., key columns or grouping columns) are evaluated against the source table. Each of the groupColumns is read, and a combined key is created. All rows from the source table are mapped to an aggregation state. Each row that has the same values for the groupColumns are mapped to the same state. Rows that have different values for the groupColumns are mapped to different states. In the case where no groupColumns are specified, all rows map to a single aggregation state. Processing continues to 406.

At 406, a new table is created containing all the distinct rows identified at 404. Processing continues to 408.

At 408, one or more aggregation columns (or result columns) is added to the new table based on the aggregation operation and/or the zero or more non-grouping (or non-key) columns. For example, a new column can be added for each non-grouping (or non-key) column with the same name and containing a collection (e.g., an array) with one element for each row that maps to the given distinct grouping key (set of grouping column values) as shown, for example, in FIGS. 5B and 5C. As can be seen in the result shown in FIG. 5C, the result columns A and B produce collections of elements from the original table. Such collections can maintain the ordering (e.g., row ordering) from the original or source table. In some examples there is not an exact match in the cardinality. For example, not all of the columns will be brought over for some operations (e.g., the “Count” operation only produces one additional result column). In other examples, several result columns can be created from one input column (e.g., the combination aggregation). For example, the Sum, Min, and Max from one input column can be applied in the same aggregation operation (e.g., an combination aggregation such as, for example, that shown in FIGS. 6A-6B in which the Min and Max are applied to the same “Timestamp” input column and the Sum and Average are applied to the same “Size” input column). In another example, two or more non-key columns of the source table can be used to create fewer result columns (e.g., a value column and a weight column of a source table can be used by a Weighted Average operation producing a single result column).

In some implementations, the array columns added at 408 (corresponding to columns not addressed explicitly in the grouping columns) are not actually instantiated in memory. In some such implementations, aggregation operations (e.g., the by( ) operation) only computes a mapping of their value in the original table, and this mapping is shared among all non-grouping (or non-key) columns, thereby reducing memory usage for aggregation operations. In some implementations, the appropriate result mapping can be determined from a subset of columns (e.g., only the key columns), and then applied to multiple result columns, thereby improving memory utilization and processor consumption. Consequently, it is relatively cheap to do something like Table1.by(“GroupCol”), even for a very large table “Table1”. In some implementations, the cost of a by(groupColumns) operation is comparable to select(groupColumns) in terms of execution time, and the storage overhead is driven by the storage requirements of the groupColumns.

In some implementations, dedicated aggregators (or predefined aggregation functions) can apply a given standard operation to all non-key columns such as, for example, those listed in Table 1 below.

TABLE 1
Predefined Aggregation Functions
Predefined Aggregation Function Syntax Description
Table sumBy(String . . . groupByColumns) Group by key columns, then calculate the
                                 sum of all other columns
Table avgBy(String . . . groupByColumns) Group by key columns, then calculate the
                                 average of all other columns
Table stdBy(String . . . groupByColumns) Group by key columns, then calculate the
                                 standard deviation of all other columns
Table varBy(String . . . groupByColumns) Group by key columns, then calculate the
                                 variance of all other columns
Table medianBy(String . . .      Group by key columns, then calculate the
groupByColumns)                  median of all other columns
Table lastBy(String . . . groupByColumns) Group by key columns, then return the last
                                 one. This generates a table with the last row
                                 with each distinct “groupbyColumns” value.
Table firstBy(String . . . groupByColumns) Group by key columns, then return the first
                                 one. This generates a table with the first row
                                 with each distinct “groupbyColumns” value.
Table minBy(String . . . groupByColumns) Group by key columns, then calculate the
                                 minimum of all other columns
Table maxBy(String . . . groupByColumns) Group by key columns, then calculate the
                                 maximum of all other columns
Table countBy(String countColumn, Group by key columns, then count number of
String . . . groupByColumns)     rows in each group.
Table countBy(String countColumn) Give the total number of values in the
                                 column, equivalent to
                                 by( ).count(countColumn).

Examples of predefined aggregation functions are shown in FIGS. 5E-H. FIG. 5E shows a result of a lastBy( ) operation in accordance with some implementations. FIG. 5F shows a result of avgBy( ) operation in accordance with some implementations. FIG. 5G shows a result of sumBy( ) operation in accordance with some implementations. FIGS. 5F and 5G show the result of avgBy( ) and sumBy( ) operations, respectively, applied to results of applying the view( ) operation on the source table shown in FIG. 5A.

In some implementations, a combined aggregation can be used to perform multiple aggregations together. In some such implementations, the following syntax is used: “resultTable=sourceTable.by(AggCombo(ComboBy . . . Aggregators), String . . . groupByColumns);”, where each aggregation is defined by a “ComboBy” object that can, for example, be created using methods of a “ComboAggregateFactory” class. The ComboAggregateFactory class can, for example, provide the following methods: ComboBy AggMin(String . . . matchPairs); ComboBy AggMax(String . . . matchPairs); ComboBy AggSum(String . . . matchPairs); ComboBy AggVar(String . . . matchPairs); ComboBy AggAvg(String . . . matchPairs); ComboBy AggStd(String . . . matchPairs); ComboBy AggFirst(String . . . matchPairs); ComboBy AggLast(String . . . matchPairs); ComboBy AggArray(String . . . matchPairs); and ComboBy AggCount(String resultColumn).

The source column names for the calculations and the names for their respective results in the new table can be provided as arguments to the ComboAggregateFactory methods. For example, if calculating both the earliest timestamp and the lowest price, the appropriate ComboBy would be created by calling: AggMin(“Timestamp”, “Price”). The resulting column can be given a different name than the source column: AggMin(“EarliestTimestamp=Timestamp”, “LowestPrice=Price”).

This syntax can apply to all of the methods except AggCount( ). Because AggCount( ) determines the count based on the number of rows rather than any particular source column, the only argument to AggCount( ) is the desired result column name: AggCount(“Count”) or AggCount(“Total”).

Passing multiple ComboBy objects can provide a concise and expressive syntax for aggregation. An example illustrating the use of multiple ComboBy objects is shown in FIGS. 6A-C and described below. In this example, table1 shown in FIG. 6A contains records of option trades in companies ABC and XYZ. Table1 can be used to determine the first and last trade times, the total size traded, and the average trade size, which can be performed by calculating minima, maxima, sums, and averages—four different types of aggregations. It would be inefficient to perform minBy( ), maxBy( ), sumBy( ), and avgBy( ) in four separate steps, creating four separate tables, then to join the results. Instead, the desired result can be produced more efficiently, and with simpler, more compact code, using the example combined aggregation syntax shown in FIG. 6B. The result of processing the combined aggregation code shown in FIG. 6B is shown in FIG. 6C.

In some implementations, a “byExternal( )” operation can divide a single table into multiple tables, which can be accessed through a “TableMap”. For example, a large table containing values for many USyms can be divided into a TableMap using: curveLog=db.getLastByTable(“SystemEQ”, “CurveLog”).where(“Date=currentDateNy( )”); and curveLogByUsym=curveLog.byExternal(“USym”). To retrieve an individual table from the map: spy=curveLogByUsym.get(“SPY”). The set of keys can be obtained using the TableMap getKeySet( ) method.

In some implementations, a byExternal(“USym”) operation is performed to divide a table for a one-click action from a console. In another example, some use cases may include dividing a table into smaller pieces, each requiring a distinct where clause (e.g., tableA=table.where(“Value=‘A’”); and tableB=table.where(“Value=‘B’”)). Using byExternal(“Value”) eliminates the need to have one pass over the data for each of the values of interest.

In some implementations, user--defined aggregators can include an “applyToAllBy( )” operation having the following syntax: Table applyToAllBy(String formulaColumn, String . . . groupByColumns). The groupByColumns groups data by key columns and applies formulaColumn to each of the non-key columns. The formulaColumn represents the formula applied to each column, and uses the parameter, for example, “each” to refer to each column to which it is being applied, as shown, for example, in FIG. 5H.

Although not explicitly shown in FIG. 4, it will be appreciated that, in some implementations, the result of an aggregation operation (e.g., the tables shown in FIGS. 5B-5H, 6C, and 8B) can be automatically updated as its source table (e.g., FIGS. 5A, 6A, and 8A) (and/or any sources of the source table) is updated as shown in FIGS. 7A, 7B, 8A-8C, 9A-9E, and 12, and described herein.

FIGS. 7A and 7B show data source definitions and a corresponding directed acyclic graph query (DAG) in accordance with some implementations. In FIG. 7A, example code 700 defines the data sources as tables (t1 and t2). From the code 700 for the data sources, DAG 702 can be generated as shown by the graph in FIG. 7B. DAG 702 in FIG. 7B shows dependencies between the nodes, which correspond to table data sources.

Although the DAG in FIG. 7B includes only two nodes, DAGs can be generated with more nodes in various configurations. For example, FIGS. 14A and 14B also show data source definitions and a corresponding directed acyclic graph (DAG) in accordance with some implementations. In FIG. 14A, example code defines the data sources as tables (t1, t3, t4, and t5), where t1 and t3 are primary data sources. From the code for the data sources, a DAG can be generated as shown by the graph in FIG. 14B. The DAG in FIG. 14B shows dependencies between the nodes, which correspond to table relationships defined in FIG. 14A.

Data sources can include market data (e.g., data received via multicast distribution mechanism or through a tailer), system generated data, historical data, user input data from the remote user table server, tables programmatically generated in-memory, or something further downstream in the DAG. In general, anything represented in the data system as a table and which can refresh itself/provide data can be a data source. Also, data sources can include non-table data structures which update, for example, mathematical data structures. As shown in FIG. 7A, t2=t1.sumBy(“GroupCol”), where this creates a summation aggregation of table t1 as a new table t2. The table t2 would then get updated when t1 changes as shown, for example, in FIGS. 9A-9E and 12. Similarly, minimum, maximum, variance, average, standard deviation, first, last, by, etc. aggregations can be supported, as shown, for example, in FIG. 14B, t5=t4.stdBy(“GroupCol”), where this creates a standard deviation aggregation of table t4 as a new table t5.

In some implementations, code can be converted into the in-memory data structures holding the DAG. For example, the source code of FIG. 7A gets converted into the DAG data structure in memory. The DAG connectivity can change by executing code. For example, assume a set of code CODE1 is executed. CODE1 leads to a DAG1 being created. Data can be processed through DAG1, leading to table updates. Now assume that the user wants to compute a few more tables. The user can run a few more lines of code CODE2, which use variables computed in CODE1. The execution of CODE2 leads to a change in the DAG. As a simple example, assume that the first 3 lines in FIG. 14A are executed. The user could come along later and execute line 4, which would modify the DAG data structure (i.e., adding t5). Also, some implementations can permit other programs to listen to changes from a node representing a data object (e.g., table or non-table object).

In some implementations, when a table changes, an application programming interface (API) can specify, for example, rows where add, modify, delete, or reindex (AMDR) changes were made. A reindex is a change in which a row is moved but the value contained in the row is not modified. The API can also provide a mechanism to obtain a value prior to the most recent change. When the DAG is processed during the refresh, the AMDR info on “upstream” data objects (e.g., tables, etc.) or nodes can be used to compute changes in “downstream” data objects or nodes. In some implementations, the entire DAG can be processed during the refresh cycle.

In general, a DAG can be comprised of a) dynamic nodes (DN); b) static nodes (SN); and c) internal nodes (IN) that can include nodes with DN and/or SN and/or IN as inputs.

DNs are nodes of the graph that can change. For example, DN can be data sources that update as new data comes in. DN could also be timers that trigger an event based on time intervals. In other examples, DN could also be MySQL monitors, specialized filtering criteria (e.g., update a “where” filter only when a certain event happens). Because these nodes are “sources”, they may occur as root nodes in the DAG. At the most fundamental level, DN are root DAG nodes which change (e.g., are “alive”).

SNs are nodes of the DAG that do not change. For example, historical data does not change. IN are interior nodes of the DAG. The state of an IN can be defined by its inputs, which can be DN, SN, and or IN. If all of the IN inputs are “static”, the IN will be static. If one or more of the IN inputs is “dynamic”, the IN will be dynamic. IN can be tables or other data structures. For example, a “listener IN” can permit code to listen to a node of the DAG. A listener node or associated listener monitoring code can place (or “fire”) additional events (or notifications) into a priority queue of a DAG.

In general, a DAG can be composed of static and/or dynamic subgraphs. In some implementations, update processing occurs on dynamic subgraphs (because static subgraphs are not changing). In some such implementations, only dynamic nodes are in the DataMonitor loop. For Tables, change notification messages such as, for example, AMDR messages can be used for communication within the DAG.

When query code is executed, the DAG is created or modified. As part of this process, the system records the order in which the DAG nodes were constructed in. This “construction ordering” can be used to determine the order that nodes are processed in the DAG.

For example, consider:

a=db.i( . . . ), where a is a dynamic node (or DN)

b=a.where(“A=1”)

c=b.where(“A=2”)

d=c.join(b)

Assume (a) has changes to be processed during a refresh cycle. The order of processing will be (a), (b), (c), and then (d).

When (d) is processed, it will process input changes from both (b) and (c) before creating AMDRs notification messages for (d). This ordering prevents (d) from creating more than one set of AMDRs per input change, and it can help ensure that all AMDRs are consistent with all data being processed for the clock cycle. If this ordering were not in place, it may be possible to get multiple ticks per cycle and some of the data can be inconsistent. Also, the ordering can help ensure that joins produce consistent results.

In some examples, a single data source is used more than once (i.e., has two or more child nodes in the DAG).

It will be appreciated that the implementations discussed above can use any update message format and are not limited to AMDR messages.

In some implementations, refresh processing of a DAG such as those shown in FIGS. 7B and 14B can be performed as disclosed in U.S. patent application Ser. No. 15/154,975, entitled “COMPUTER DATA SYSTEM DATA SOURCE REFRESHING USING AN UPDATE PROPAGATION GRAPH” (Attorney Docket No. W1.4-10058) and filed on May 14, 2016 (hereinafter the '975 application), which is hereby incorporated by reference herein in its entirety as if fully set forth herein. For example, refresh processing of the DAG can be performed in accordance with the data source refresh processes disclosed by FIG. 6 and the specification of the '975 application, where the notifications delivered at 614 include the AMDR notifications shown in FIGS. 9A-9E of the present disclosure and described below.

FIG. 8A shows a data source table (t1 defined in FIG. 7A) in accordance with some implementations. Table t1 includes three columns (GroupCol 402, DataCol1404, and DataCol2408) and six rows including Row1410 and Row5412.

FIGS. 8B and 8C illustrate aggregation table t2 that is a result of the aggregation operation shown in FIGS. 7A-7B in accordance with some implementations. Table t2 is the result of the aggregation operation sumBy(“GroupCol”) shown in FIG. 7A. Table t2 includes three columns (GroupCol 602, DataCol1604, DataCol2606) corresponding to the columns of t1 (402, 404, and 406, respectively) and aggregates rows based on the column specified in the sumBy( ) call—the “GroupCol” column. Accordingly, table t2 includes a row for each of the distinct values of the GroupCol 402 column of t1: Row0610 for those rows of t1 with “C” in the GroupCol 402 column, Row1612 for those rows of t1 with “D” in the GroupCol 402 column, and Row2608 for those rows of t1 with “E” in the GroupCol 402 column.

In some implementations, an aggregation state is created for each row of the result of an aggregation operation as shown, for example, in FIG. 8C. FIG. 8C shows aggregation table t2 including a listener 1302 and aggregation states 1304 in accordance with some implementations. Aggregation states 1304 includes aggregation states 1306, 1308, and 1310 for Row0610, Row1612, and Row2608 of t2, respectively. Table t2 can subscribe to updates from its parent or upstream source table t1 and receive at listener 1302 AMDR messages (e.g., via the DAG) indicating updates to t1. The AMDR messages can be processed according to the method shown, for example, in FIG. 12. In some implementations listener 1302 receives AMDR messages, identifies the appropriate aggregation state(s) to handle the messages (e.g., including creating/removing aggregation states when appropriate), and calls the identified aggregation state(s) to process the updates indicated in the AMDR messages, as shown, for example, in FIGS. 9A-9E.

FIG. 9A illustrates automatic updating of an aggregation result when a row is modified in a source table in accordance with some implementations. When row 412 of t1 is modified, t2 is notified via an AMDR message. The AMDR message can be generated based on the DAG created when t2 was defined as shown, for example, in FIGS. 7A and 7B and discussed above. The DAG node for t2 can receive the AMDR message and the update for the corresponding aggregation row 608 can be determined by accessing at most only a subset of rows of the source table that had previously been aggregated for aggregation row 608 (e.g., determined without having to access the other rows of t1 that had previously been aggregated for aggregation row 608) according to, for example, the method shown in FIG. 12. The term subset is meant to mean less than all of the entire set.

In some implementations, processing the update can include determining that the key (i.e., value of the grouping column 402) of row 412 (i.e., “E”) has not changed (by comparing its oldvalue and newvalue for the current update/refresh/clock cycle) and then identifying the aggregation state(s) associated with that key (e.g., aggregation state 1310 shown in FIG. 8C). In some implementations, different aggregation functions can be applied to different columns and different aggregation states can be created for the different aggregation functions. In some implementations, the same aggregation function can be applied to multiple columns and one aggregation state or multiple aggregation states can be created for the multiple columns. The associated aggregation state(s) can process the row modification by accessing a small subset of rows of the source table aggregated by the aggregation state (e.g., without accessing the values of the other rows of t1 that had previously been aggregated by the aggregation state). For example, an aggregation state for the summation aggregation shown can store a running sum for each column being aggregated. Such an aggregation state can process the row modification by, for each column being aggregated, subtracting the oldvalue of the modified row's corresponding column from, and adding the newvalue of the modified row's corresponding column to, the running sum (i.e., the value before processing the modification to row 412 of t1).

For column 606 of aggregation row 608, the current aggregation value is the result of the sum of (N+Q), and the summation aggregation state associated with that column/row can determine an updated aggregation value by subtracting the corresponding oldvalue(Q) from (N+Q) and adding the corresponding newvalue(R). It can be determined (e.g., by the aggregation state) that the aggregation value has changed and an AMDR notification indicating the change can be generated for any child nodes (or subscribers) of t2.

For column 604 of aggregation row 608, the current aggregation value is the result of (H+K), and the summation aggregation state can determine the updated aggregation value by subtracting the corresponding oldvalue(K) and adding the corresponding newvalue(K).

FIG. 9B illustrates automatic updating of an aggregation result when a row is modified in a source table and the modification includes a change to the key in accordance with some implementations. When row 412 of t1 is modified as shown in FIG. 9B, t2 is notified via an AMDR message.

In some implementations, processing the update can include determining that the key (i.e., value of the grouping column 402) of row 412 has changed from “E” to “C” (by comparing its oldvalue and newvalue for the current update/refresh/clock cycle) and identifying the aggregation state(s) associated with those key (e.g., aggregation states 1306 and 1310 shown in FIG. 8C). The associated aggregation state(s) can process the row modification by accessing a small subset of rows of the source table aggregated by the aggregation state (e.g., without accessing the values of other rows of t1 that had previously been aggregated by the aggregation state). The aggregation state associated with the key's oldvalue can process the update by, for each aggregated column, removing the oldvalue of the column in row 412 from the aggregation value. The aggregation state associated with the key's newvalue can process the update by, for each aggregated column, adding the newvalue of the column in row 412 to the aggregation value.

For column 604 of aggregation row 610, the current aggregation value is the result of the sum of (F+I+J), and the summation aggregation state associated with that column/row can determine an updated aggregation value by adding the corresponding newvalue(K) to (F+I+J). For column 606 of aggregation row 610, the current aggregation value is the result of (L+O+P), and the summation aggregation state can determine the updated aggregation value by adding the corresponding newvalue(T). It can be determined (e.g., by the aggregation state) that the aggregation value(s) of aggregation row 610 has changed and an AMDR notification indicating the change can be generated for any child nodes (or subscribers) of t2.

For column 604 of aggregation row 608, the current aggregation value is the result of the sum of (H+K), and the summation aggregation state associated with that column/row can determine an updated aggregation value by subtracting the corresponding oldvalue(K) from (H+K). For column 606 of aggregation row 608, the current aggregation value is the result of (N+Q), and the summation aggregation state can determine the updated aggregation value by subtracting the corresponding oldvalue(Q). It can be determined (e.g., by the aggregation state) that the aggregation value(s) of aggregation row 608 has changed and an AMDR notification indicating the change can be generated for any child nodes (or subscribers) of t2.

FIG. 9C illustrates automatic updating of an aggregation result when a row is modified in a source table and the modification includes a change to the key in accordance with some implementations. When row 410 of t1 is modified as shown in FIG. 9C, t2 is notified via an AMDR message.

In some implementations, processing the update can include determining that the key (i.e., value of the grouping column 402) of row 410 has changed from “D” to “C” (by comparing its oldvalue and newvalue for the current update/refresh/clock cycle) and identifying the aggregation state(s) associated with those key (e.g., aggregation states 1306 and 1308 shown in FIG. 8C). The associated aggregation state(s) can process the row modification by accessing a small subset of rows of the source table aggregated by the aggregation state (e.g., without accessing the values of other rows of t1 that had previously been aggregated by the aggregation state). The aggregation state associated with the key's oldvalue can process the update by, for each aggregated column, removing the oldvalue of the column in row 410 from the aggregation value. The aggregation state associated with the key's newvalue can process the update by, for each aggregated column, adding the newvalue of the column in row 410 to the aggregation value.

For column 604 of aggregation row 610, the current aggregation value is the result of the sum of (F+I+J), and the summation aggregation state associated with that column/row can determine an updated aggregation value by adding the corresponding newvalue(S) to (F+I+J). For column 606 of aggregation row 610, the current aggregation value is the result of (L+O+P), and the summation aggregation state can determine the updated aggregation value by adding the corresponding newvalue(T). It can be determined (e.g., by the aggregation state) that the aggregation value(s) of aggregation row 610 has changed and an AMDR notification indicating the change can be generated for any child nodes (or subscribers) of t2.

For column 604 of aggregation row 612, the current aggregation value is the result of the sum of (G), and the summation aggregation state associated with that column/row can determine an updated aggregation value by subtracting the corresponding oldvalue(G) from (G). For column 606 of aggregation row 612, the current aggregation value is the result of (M), and the summation aggregation state can determine the updated aggregation value by subtracting the corresponding oldvalue(M). It can be determined (e.g., by the aggregation state) that aggregation row 608 now contains zero entries, and the aggregation row 608 can be removed from t2 and an AMDR notification indicating the change can be generated for any child nodes (or subscribers) of t2.

FIG. 9D illustrates automatic updating of an aggregation result when a row is added to a source table in accordance with some implementations. When row 906 is added to t1 is added as shown in FIG. 9D, t2 is notified via an AMDR message.

In some implementations, processing the update can include identifying the aggregation state(s) associated with the key (i.e., value of the grouping column 402) of row 906 (e.g., aggregation state 1310 shown in FIG. 8C). The associated aggregation state(s) can process the row modification by accessing a small subset of rows of the source table aggregated by the aggregation state (e.g., without accessing the values of other rows of t1 that had previously been aggregated by the aggregation state). The associated aggregation state(s) can process the update by, for each aggregated column, adding the newvalue of the column in row 906 to the aggregation value.

For column 604 of aggregation row 608, the current aggregation value is the result of the sum of (H+K), and the summation aggregation state associated with that column/row can determine an updated aggregation value by adding the corresponding newvalue(K) to (H+K). For column 606 of aggregation row 608, the current aggregation value is the result of (N+Q), and the summation aggregation state can determine the updated aggregation value by adding the corresponding newvalue(A). It can be determined (e.g., by the aggregation state) that the aggregation value(s) of aggregation row 608 has changed and an AMDR notification indicating the change can be generated for any child nodes (or subscribers) of t2.

FIG. 9E illustrates automatic updating of an aggregation result when a row is deleted from a source table in accordance with some implementations. When row 412 is deleted from t1 is added as shown in FIG. 9E, t2 is notified via an AMDR message.

In some implementations, processing the update can include identifying the aggregation state(s) associated with the key (i.e., value of the grouping column 402) of row 412 (e.g., aggregation state 1310 shown in FIG. 8C). The associated aggregation state(s) can process the row modification by accessing a small subset of rows of the source table aggregated by the aggregation state (e.g., without accessing the values of other rows of t1 that had previously been aggregated by the aggregation state). The associated aggregation state(s) can process the update by, for each aggregated column, deleting the oldvalue of the column in row 412 from the aggregation value.

For column 604 of aggregation row 608, the current aggregation value is the result of the sum of (H+K), and the summation aggregation state associated with that column/row can determine an updated aggregation value by subtracting the corresponding oldvalue(K) from (H+K). For column 606 of aggregation row 608, the current aggregation value is the result of (N+Q), and the summation aggregation state can determine the updated aggregation value by subtracting the corresponding oldvalue(Q). It can be determined (e.g., by the aggregation state) that the aggregation value(s) of aggregation row 608 has changed and an AMDR notification indicating the change can be generated for any child nodes (or subscribers) of t2.

FIGS. 10 and 11 are diagrams of example computer data system 100 showing an example data distribution configuration in accordance with some implementations. Source table t1 can be stored in periodic data 122 and aggregation result table t2 can be stored at remote query processor 134, as shown in FIG. 10, and/or at table user data 144, as shown in FIG. 11.

FIG. 13 is a diagram of example computer data system 100 showing an example data distribution configuration in accordance with some implementations. Source table t1 can be stored in periodic data 122, source table t3 can be stored at long-term data 140, source table t4 can be stored at remote query processor 134, and aggregation result table t5 can be stored at user table data 144.

FIG. 12 is a flowchart of an example aggregation operation in accordance with some implementations. Processing begins at 1202.

At 1202, Table B is created by calling a method on Table A (e.g., by( ), sumBy( ), or other aggregation operation). Processing continues to 1204.

At 1204, Table B creates a listener object, which refers to Table A and subscribes to AMDR messages. Table A can be a source table that generates an AMDR, or in response to an update from one of its parents, generates an AMDR. Processing continues to 1206.

At 1206, Table B's listener receives the AMDR message(s). Processing continues to 1208.

At 1208, for each deleted row, the previous value of the key columns is read, and is used to identify which aggregation state (or states) that the row corresponded to. The row is marked for deletion from that aggregation state (or states). Processing continues to 1210.

At 1210, for each modified row, the previous value of the key columns is read, and the current value of the key columns is read. Processing continues to 1212.

At 1212, the previous (i.e., oldvalue) and current (i.e., newvalue) values of the key columns are compared. If they are identical, processing continues to 1214. Otherwise, processing continues to 1216.

At 1214, the aggregation state is identified and the row is marked for modification (e.g., FIG. 9A). Processing continues to 1218.

At 1216, the old aggregation state is identified and the row is marked for deletion, and the new aggregation state is identified and the row is marked for addition (e.g., FIG. 9B). Processing continues to 1218.

At 1218, for each added row, the current value of the key columns is read and the aggregation state is identified. The row is marked for addition (e.g., FIG. 9D). Processing continues to 1220.

At 1220, the marked deleted/modified/added rows are delivered to each aggregation state. Processing continues to 1222.

At 1222, each aggregation state uses the marked deleted rows, modified rows, and added rows to compute a new value for its output. In some implementations, the aggregation states can use the clock cycle oldvalues and/or newvalues to compute the new value for its output as shown, for example, in FIGS. 9A-9E. Processing continues to 1224.

At 1224, if the aggregation state detects a change to its output value, then it marks its row as added, modified, or deleted. Processing continues to 1226.

At 1226, the. query engine iterates over the referenced aggregation states and produces a set of AMDR update for any downstream tables (e.g., tables that have subscribed to receive updates from Table B.

Some implementations include a method for maximizing memory and processing efficiencies in a computer system. In some such implementations, the method can include receiving an aggregation query task (e.g., at a remote query processor such as RQP 132) to create a second table object comprising an updateable aggregation of rows from a first table object. The data of the first table object can be stored in a computer storage and the second table object can be creating and stored in a computer memory separate from the computer storage, in some implementations.

The method can include creating a plurality of aggregation states for a plurality of aggregation rows of the second table object, each of the plurality of aggregation rows representing an aggregation of one or more rows of the first table object, each of the plurality of aggregation states being configured to determine an aggregated value for a respective one of the plurality of aggregation rows by executing an aggregation formula. The aggregated values can comprise a collection (e.g., an array) for each of one or more aggregated columns of each row of the second table object; and the collection can comprise references to data of the first table object.

The method can further include storing determined aggregated values in the computer memory. In some implementations, the aggregated values are stored in the computer memory separate from the computer storage.

The method can include creating and starting a listener for the second table object.

The method can include subscribing to receive at the listener an indication of an update to the first table object. The method can further include, responsive to receiving the indication at the listener, selecting one or more of the aggregation states to update the second table object and accessing at most only a subset of rows of the first table object, thereby eliminating the use of repeat processing time for re-executing the aggregation formula and providing faster access to the plurality of aggregation rows of the second table object.

In some implementations, the method can include detecting a change to an output of an aggregation state of the plurality of aggregation states, and transmitting, responsive to the detecting, a change notification (e.g., an AMDR) to a subscriber of the second table object.

It will be appreciated that the modules, processes, systems, and sections described above can be implemented in hardware, hardware programmed by software, software instructions stored on a nontransitory computer readable medium or a combination of the above. A system as described above, for example, can include a processor configured to execute a sequence of programmed instructions stored on a nontransitory computer readable medium. For example, the processor can include, but not be limited to, a personal computer or workstation or other such computing system that includes a processor, microprocessor, microcontroller device, or is comprised of control logic including integrated circuits such as, for example, an Application Specific Integrated Circuit (ASIC), a field programmable gate array (FPGA) or the like. The instructions can be compiled/interpreted from source code instructions provided in accordance with a programming language such as Java, C, C++, C#.net, assembly or the like. The instructions can also comprise code and data objects provided in accordance with, for example, the Visual Basic™ language, a specialized database query language, or another structured, object-oriented or other programming language. The sequence of programmed instructions, or programmable logic device configuration software, and data associated therewith can be stored in a nontransitory computer-readable medium such as a computer memory or storage device which may be any suitable memory apparatus, such as, but not limited to ROM, PROM, EEPROM, RAM, flash memory, disk drive and the like.

Furthermore, the modules, processes systems, and sections can be implemented as a single processor or as a distributed processor. Further, it should be appreciated that the steps mentioned above may be performed on a single or distributed processor (single and/or multi-core, or cloud computing system). Also, the processes, system components, modules, and sub-modules described in the various figures of and for embodiments above may be distributed across multiple computers or systems or may be co-located in a single processor or system. Example structural embodiment alternatives suitable for implementing the modules, sections, systems, means, or processes described herein are provided below.

The modules, processors or systems described above can be implemented as a programmed general purpose computer, an electronic device programmed with microcode, a hard-wired analog logic circuit, software stored on a computer-readable medium or signal, an optical computing device, a networked system of electronic and/or optical devices, a special purpose computing device, an integrated circuit device, a semiconductor chip, and/or a software module or object stored on a computer-readable medium or signal, for example.

Embodiments of the method and system (or their sub-components or modules), may be implemented on a general-purpose computer, a special-purpose computer, a programmed microprocessor or microcontroller and peripheral integrated circuit element, an ASIC or other integrated circuit, a digital signal processor, a hardwired electronic or logic circuit such as a discrete element circuit, a programmed logic circuit such as a PLD, PLA, FPGA, PAL, or the like. In general, any processor capable of implementing the functions or steps described herein can be used to implement embodiments of the method, system, or a computer program product (software program stored on a nontransitory computer readable medium).

Furthermore, embodiments of the disclosed method, system, and computer program product (or software instructions stored on a nontransitory computer readable medium) may be readily implemented, fully or partially, in software using, for example, object or object-oriented software development environments that provide portable source code that can be used on a variety of computer platforms. Alternatively, embodiments of the disclosed method, system, and computer program product can be implemented partially or fully in hardware using, for example, standard logic circuits or a VLSI design. Other hardware or software can be used to implement embodiments depending on the speed and/or efficiency requirements of the systems, the particular function, and/or particular software or hardware system, microprocessor, or microcomputer being utilized. Embodiments of the method, system, and computer program product can be implemented in hardware and/or software using any known or later developed systems or structures, devices and/or software by those of ordinary skill in the applicable art from the function description provided herein and with a general basic knowledge of the software engineering and computer networking arts.

Moreover, embodiments of the disclosed method, system, and computer readable media (or computer program product) can be implemented in software executed on a programmed general purpose computer, a special purpose computer, a microprocessor, or the like.

It is, therefore, apparent that there is provided, in accordance with the various embodiments disclosed herein, methods, systems and computer readable media for memory-efficient and processor-efficient aggregation and data source refreshing.

Application Ser. No. 15/154,974, entitled “DATA PARTITIONING AND ORDERING” (Attorney Docket No. W1.1-10057) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,975, entitled “COMPUTER DATA SYSTEM DATA SOURCE REFRESHING USING AN UPDATE PROPAGATION GRAPH” (Attorney Docket No. W1.4-10058) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,979, entitled “COMPUTER DATA SYSTEM POSITION-INDEX MAPPING” (Attorney Docket No. W1.5-10083) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,980, entitled “SYSTEM PERFORMANCE LOGGING OF COMPLEX REMOTE QUERY PROCESSOR QUERY OPERATIONS” (Attorney Docket No. W1.6-10074) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,983, entitled “DISTRIBUTED AND OPTIMIZED GARBAGE COLLECTION OF REMOTE AND EXPORTED TABLE HANDLE LINKS TO UPDATE PROPAGATION GRAPH NODES” (Attorney Docket No. W1.8-10085) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,984, entitled “COMPUTER DATA SYSTEM CURRENT ROW POSITION QUERY LANGUAGE CONSTRUCT AND ARRAY PROCESSING QUERY LANGUAGE CONSTRUCTS” (Attorney Docket No. W2.1-10060) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,985, entitled “PARSING AND COMPILING DATA SYSTEM QUERIES” (Attorney Docket No. W2.2-10062) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,987, entitled “DYNAMIC FILTER PROCESSING” (Attorney Docket No. W2.4-10075) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,988, entitled “DYNAMIC JOIN PROCESSING USING REAL-TIME MERGED NOTIFICATION LISTENER” (Attorney Docket No. W2.6-10076) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,990, entitled “DYNAMIC TABLE INDEX MAPPING” (Attorney Docket No. W2.7-10077) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,991, entitled “QUERY TASK PROCESSING BASED ON MEMORY ALLOCATION AND PERFORMANCE CRITERIA” (Attorney Docket No. W2.8-10094) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,993, entitled “A MEMORY-EFFICIENT COMPUTER SYSTEM FOR DYNAMIC UPDATING OF JOIN PROCESSING” (Attorney Docket No. W2.9-10107) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,995, entitled “QUERY DISPATCH AND EXECUTION ARCHITECTURE” (Attorney Docket No. W3.1-10061) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,996, entitled “COMPUTER DATA DISTRIBUTION ARCHITECTURE” (Attorney Docket No. W3.2-10087) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,997, entitled “DYNAMIC UPDATING OF QUERY RESULT DISPLAYS” (Attorney Docket No. W3.3-10059) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,998, entitled “DYNAMIC CODE LOADING” (Attorney Docket No. W3.4-10065) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/154,999, entitled “IMPORTATION, PRESENTATION, AND PERSISTENT STORAGE OF DATA” (Attorney Docket No. W3.5-10088) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/155,001, entitled “COMPUTER DATA DISTRIBUTION ARCHITECTURE” (Attorney Docket No. W3.7-10079) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/155,005, entitled “PERSISTENT QUERY DISPATCH AND EXECUTION ARCHITECTURE” (Attorney Docket No. W4.2-10089) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/155,006, entitled “SINGLE INPUT GRAPHICAL USER INTERFACE CONTROL ELEMENT AND METHOD” (Attorney Docket No. W4.3-10063) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/155,007, entitled “GRAPHICAL USER INTERFACE DISPLAY EFFECTS FOR A COMPUTER DISPLAY SCREEN” (Attorney Docket No. W4.4-10090) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/155,009, entitled “COMPUTER ASSISTED COMPLETION OF HYPERLINK COMMAND SEGMENTS” (Attorney Docket No. W4.5-10091) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/155,010, entitled “HISTORICAL DATA REPLAY UTILIZING A COMPUTER SYSTEM” (Attorney Docket No. W5.1-10080) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/155,011, entitled “DATA STORE ACCESS PERMISSION SYSTEM WITH INTERLEAVED APPLICATION OF DEFERRED ACCESS CONTROL FILTERS” (Attorney Docket No. W6.1-10081) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

Application Ser. No. 15/155,012, entitled “REMOTE DATA OBJECT PUBLISHING/SUBSCRIBING SYSTEM HAVING A MULTICAST KEY-VALUE PROTOCOL” (Attorney Docket No. W7.2-10064) and filed in the United States Patent and Trademark Office on May 14, 2016, is hereby incorporated by reference herein in its entirety as if fully set forth herein.

While the disclosed subject matter has been described in conjunction with a number of embodiments, it is evident that many alternatives, modifications and variations would be, or are, apparent to those of ordinary skill in the applicable arts. Accordingly, Applicants intend to embrace all such alternatives, modifications, equivalents and variations that are within the spirit and scope of the disclosed subject matter.

Claims

1. A system for maximizing memory and processing efficiencies in a computer system, the system comprising: one or more processors;computer readable storage coupled to the one or more processors, the computer readable storage having stored thereon instructions that, when executed by the one or more processors, cause the one or more processors to perform operations including: sending a digital request from a client computer to a remote query processor on a query server computer;creating and storing, in a computer storage separate from the computer readable storage, data in a plurality of column sources;creating and storing in a computer memory a first table object index mapping data in the plurality of column sources to a first table object, the computer memory having faster access time than the computer storage separate from the computer readable storage;at the remote query processor, providing remote operations including: an aggregation query operation, the aggregation query operation comprising: receiving an aggregation query task to create a second table object comprising an updateable aggregation of rows from the first table object;creating and storing in the computer memory separate from the computer storage, the second table object;creating a plurality of aggregation states for a plurality of aggregation rows of the second table object, each of the plurality of aggregation rows representing an aggregation of one or more rows of the first table object, each of the plurality of aggregation states being configured to determine an aggregated value for a respective one of the plurality of aggregation rows by executing an aggregation formula;storing determined aggregated values in the computer memory separate from the computer storage;creating and starting a listener for the second table object; andsubscribing to receive at the listener an indication of an update to the first table object and, responsive to receiving the indication at the listener, selecting one or more of the aggregation states to update the second table object and accessing a subset of rows of the first table object, wherein the subset of rows is smaller than the first table object,the computer storage separate from the computer readable storage being remote from the remote query processor and being accessed by the remote query processor via a network, andthe computer memory being local to the remote query processor.

2. The system of claim 1, wherein the aggregated values comprise a collection for each aggregated column of each row of the second table object.

3. The system of claim 2, wherein a first collection of a first aggregated column of a first row of the second table object is a first ordered collection;wherein a second collection of a second aggregated column of the first row of the second table object is a second ordered collection; andwherein the first ordered collection and the second ordered collection have a same ordering.

4. The system of claim 2, wherein the collection comprises a reference to data of the first table object.

5. The system of claim 1, wherein the aggregation function determines a summation and the aggregated values comprise a running sum.

6. The system of claim 1, wherein the aggregation function determines a standard deviation and the aggregated values comprise a running sum, a running sum of squares, and a count.

7. A system for maximizing memory and processing efficiencies in a computer system, the system comprising: one or more processors;computer readable storage coupled to the one or more processors, the computer readable storage having stored thereon instructions that, when executed by the one or more processors, cause the one or more processors to perform operations including:storing a second table object representing an updateable aggregation of rows from a first table object;creating a plurality of aggregation states for a plurality of aggregation rows of the second table object, each of the plurality of aggregation rows representing an aggregation of one or more rows of the first table object, each of the plurality of aggregation states being configured to determine an aggregated value for a respective one of the plurality of aggregation rows by executing an aggregation formula;storing determined aggregated values in the computer memory separate from the computer readable storage;receiving an indication of an update to the first table object; andresponsive to receiving, selecting one or more of the aggregation states to update the second table object and accessing a subset of rows of the first table object.

8. The system of claim 7, wherein the aggregated values comprise a collection for each aggregated column of each row of the second table object.

9. The system of claim 8, wherein a first collection of a first aggregated column of a first row of the second table object is a first ordered collection;wherein a second collection of a second aggregated column of the first row of the second table object is a second ordered collection; andwherein the first ordered collection and the second ordered collection have a same ordering.

10. The system of claim 8, wherein the collection comprises a reference to data of the first table object.

11. The system of claim 8, the operations further comprising: determining a mapping of a subset of columns and applying the mapping to one or more other columns to create one or more of the collections for the aggregated columns of the second table object.

12. The system of claim 7, wherein the aggregation function determines a weighted average.

13. The system of claim 7, wherein a combination aggregation is determined for a row of the second table object by two or more of the plurality of aggregation states.

14. The system of claim 7, wherein data of the first table object is stored in a computer storage separate from the computer readable storage that is remote from a computer memory separate from the computer readable storage and the computer storage and is accessed via a network, andwherein the computer memory separate from the computer readable storage is local to the one or more processors.

15. A nontransitory computer readable medium having stored thereon software instructions that, when executed by one or more processors, cause the one or more processors to perform operations including: storing, in a computer memory separate from the nontransitory computer readable medium, a second table object representing an updateable aggregation of rows from a first table object;creating, at a processor coupled to the computer memory, a plurality of aggregation states for a plurality of aggregation rows of the second table object, each of the plurality of aggregation rows representing an aggregation of one or more rows of the first table object, each of the plurality of aggregation states being configured to determine an aggregated value for a respective one of the plurality of aggregation rows by executing an aggregation formula;storing determined aggregated values in the computer memory;receiving an indication of an update to the first table object; andresponsive to receiving, selecting one or more of the aggregation states to update the second table object and accessing a subset of rows of the first table object, wherein the subset of rows is smaller than the first table object.

16. The nontransitory computer readable medium of claim 15, wherein the aggregated values comprise a collection for each aggregated column of each row of the second table object.

17. The nontransitory computer readable medium of claim 16, wherein a first collection of a first aggregated column of a first row of the second table object is a first ordered collection;wherein a second collection of a second aggregated column of the first row of the second table object is a second ordered collection; andwherein the first ordered collection and the second ordered collection have a same ordering.

18. The nontransitory computer readable medium of claim 16, wherein the collection comprises a reference to data of the first table object.

19. The nontransitory computer readable medium of claim 16, the operations further comprising: determining a mapping of a subset of columns and applying the mapping to one or more other columns to create one or more of the collections for the aggregated columns of the second table object.

20. The nontransitory computer readable medium of claim 15, wherein the aggregation function determines a maximum value.

21. The nontransitory computer readable medium of claim 15, wherein data of the first table object is stored in a computer storage separate from the nontransitory computer readable medium and the computer memory that is remote from the computer memory and is accessed via a network, andwherein the computer memory is local to the one or more processors.

22. A method for maximizing memory and processing efficiencies in a computer system, the system comprising: storing, in a computer memory, a second table object representing an updateable aggregation of rows from a first table object;creating, at a processor coupled to the computer memory, a plurality of aggregation states for a plurality of aggregation rows of the second table object, each of the plurality of aggregation rows representing an aggregation of one or more rows of the first table object, each of the plurality of aggregation states being configured to determine an aggregated value for a respective one of the plurality of aggregation rows;storing determined aggregated values in the computer memory;receiving an indication of an update to the first table object; andresponsive to receiving, selecting one or more of the aggregation states to update the second table object and accessing a subset of rows of the first table object.

23. The method of claim 22, wherein the aggregated values comprise a collection for each of one or more aggregated columns of each row of the second table object.

24. The method of claim 23, wherein a first collection of a first aggregated column of a first row of the second table object is a first ordered collection;wherein a second collection of a second aggregated column of the first row of the second table object is a second ordered collection; andwherein the first ordered collection and the second ordered collection have a same ordering.

25. The method of claim 24, wherein the first ordering and the second ordering is determined based on an ordering of the first table object.

26. The method of claim 23, wherein the collection comprises a reference to data of the first table object.

27. The method of claim 23, further comprising: determining a mapping of a subset of columns and applying the mapping to one or more other columns to create one or more of the collections for the aggregated columns of the second table object.

28. The method of claim 22, further comprising: detecting a change to an output of an aggregation state of the plurality of aggregation states; andtransmitting, responsive to the detecting, a change notification to a subscriber of the second table object.

29. The method of claim 28, wherein the change notifications can include one of a data add notification, a data modify notification, a data delete notification and a data reindex notification.

30. The method of claim 22, wherein at least one of the plurality of aggregation states is configured to determine an aggregated value based on a new value and an associated previous value of the first table object.