Abstract: Embodiments relate to the orchestration of data shuffling among memory devices of a non-uniform memory access device. An aspect includes a method of orchestrated shuffling of data in a non-uniform memory access device includes running an application on a plurality of threads executing on a plurality of processing nodes and identifying data to be shuffled among the plurality of processing nodes. The method includes registering the data to be shuffled and generating a plan for orchestrating the shuffling of the data. The method further includes disabling cache coherency of cache memory associated with the processing nodes and shuffling the data among all of the memory devices upon disabling the cache coherency, the shuffling performed based on the plan for orchestrating the shuffling. The method further includes restoring the cache coherency of the cache memory based on completing the shuffling of the data among all of the memory devices.

Abstract: According to one embodiment of the present invention, a method of transferring filters for operations in a distributed system comprises generating at each source node a first filter of a first size. The first filter indicates elements of objects that need not be processed for an operation. Each source node generates one more second filters of a second size less than the first size. Each generated second filter is based on a prior generated filter with each location representing a plurality of locations from the prior generated filter. A second filter is transferred from each source node to each of destination node. The elements of the objects to remove for the operation are determined based on the transferred second filters. Embodiments of the present invention further include a system and computer program product for transferring filters in a distributed system in substantially the same manners described above.

Abstract: Embodiments of the invention relate to processing data records, and for a multi-phase partitioned data reduction. The first phase relates to processing data records and partitioning the records into a first partition of records having a common characteristic and a second partition of records that are not members of the first partition. The data records in each partition are subject to intra-partition data reduction responsive to a resource constraint. The data records in each partition are also subject to an inter-partition data reduction, also referred to as an aggregation to reduce a footprint for storing the records. Partitions and/or individual records are logically aggregated and a data reduction operation for the logical aggregation of records takes place in response to available resources.

Abstract: According to one embodiment of the present invention, a method of transferring filters for operations in a distributed system comprises generating at each source node a first filter of a first size. The first filter indicates elements of objects that need not be processed for an operation. Each source node generates one more second filters of a second size less than the first size. Each generated second filter is based on a prior generated filter with each location representing a plurality of locations from the prior generated filter. A second filter is transferred from each source node to each of destination node. The elements of the objects to remove for the operation are determined based on the transferred second filters. Embodiments of the present invention further include a system and computer program product for transferring filters in a distributed system in substantially the same manners described above.

Abstract: A system joins predicate evaluated column bitmaps having varying lengths. The system includes a column unifier for querying column values with a predicate and generating an indicator bit for each of the column values that is then joined with the respective column value. The system also includes a bitmap generator for creating a column-major linear bitmap from the column values and indicator bits. The column unifier also determines an offset between adjacent indicator bits. The system also includes a converter for multiplying the column-major linear bitmap with a multiplier to shift the indicator bits into consecutive positions in the linear bitmap.

Abstract: Embodiments relate to hierarchical database compression. An aspect includes applying a first level of a first type of compression to a first partition of a column of a database. Another aspect includes applying a second level of the first type of compression to a subset of the first partition, wherein the first level of the first type of compression comprises a first first-level dictionary and the second level of the first type of compression comprises a first second-level dictionary, and wherein a code size of the first first-level dictionary is larger than a code size of the first second-level dictionary.

Abstract: Embodiments include method, system, and computer program product for providing aggregation hierarchy that is related memory hierarchies. In one embodiment, the method includes determining capacity of a first level memory of a memory hierarchy for processing data relating to completion of an aggregation process and generating a per thread local look-up table in said first level memory upon determining said capacity. Upon the first level memory reaching capacity, a plurality of per thread partitions to store remaining data to complete the aggregation process in a second level memory of the memory hierarchy is generated such that each of said per-thread partitions includes an identical amount of data portion on each thread. The method also includes storing the per thread partitions in said second level memory and providing a single global look up table for each of the identical data portions.

Abstract: Embodiments of the invention relate to processing queries. A query operation to be performed on a table of data is translated into a series of bit level logical operations using expansion and/or saturation operations. A mask is created from the series of bit level logical operations. This mask is then simultaneously applied to multiple rows from the table of data.

Abstract: A system joins predicate evaluated column bitmaps having varying lengths. The system includes a column unifier for querying column values with a predicate and generating an indicator bit for each of the column values that is then joined with the respective column value. The system also includes a bitmap generator for creating a column-major linear bitmap from the column values and indicator bits. The column unifier also determines an offset between adjacent indicator bits. The system also includes a converter for multiplying the column-major linear bitmap with a multiplier to shift the indicator bits into consecutive positions in the linear bitmap.

Abstract: Embodiments of the invention relate to processing queries. A query operation to be performed on a table of data is translated into a series of bit level logical operations using expansion and/or saturation operations. A mask is created from the series of bit level logical operations. This mask is then simultaneously applied to multiple rows from the table of data.

Abstract: A weight is determined for each of a plurality of join predicates for a join between one or more first database objects and one or more second database objects based on a join selectivity for each of the plurality of join predicates. The plurality of join predicates are sorted based on the determined weights. The join operation is performed joining the one or more first database objects with the one or more second database objects in accordance with an order of the sorted plurality of join predicates.

Abstract: Embodiments of the invention relate to processing queries that utilize fact and/or dimension tables. In one aspect, a pre-join filtering phase precedes a star join. The necessary conditions for the pre-join filtering are considered for a given SQL query, including an estimated size of the hash table exceeding a threshold and presence of a local predicate either on the fact table or one or more dimension tables that is not a large dimension table. Once the necessary conditions are satisfied, the execution of the query exploits the pre-join filtering to build a pre-join output filter from columns of a reduced fact table that joins with each large dimension table. Thereafter, all the dimension tables and the fact table are joined in a star join while exploiting each pre-join filter.

Abstract: Embodiments relate to an immediate join spilling scheme for a database. An aspect includes receiving a command to perform a join of an inner table and an outer table in a database. Another aspect includes spilling a plurality of hash tables to a disk memory, each hash table corresponding to a respective inner table partition. Another aspect includes registering, by a first agent of a plurality of agents, a set of inner table partitions that require processing by the first agent to complete the join for a first stride of rows of the outer table. Another aspect includes, based on determining that a hash table corresponding to a registered inner table partition of the first agent has been loaded into the local memory by the second agent of the plurality of agents, performing the join of the loaded hash table and the first stride of rows of the outer table by the first agent.

Abstract: According to one embodiment of the present invention, a system with a plurality of memories, generates a first filter associated with a first memory, and generates one or more second filters each associated with a corresponding memory. Each second filter is generated with a size at least partially accommodated within the storage capacity of the associated memory. The system determines absence of items from a set by selectively probing the generated filters in an ascending order of data access speed to the associated memories and aborts further probing for an item in response to a probe indicating that the item is absent from the set. Embodiments of the present invention further include a method and computer program product for testing set membership in substantially the same manners described above.

Abstract: According to one embodiment of the present invention, a system with a plurality of memories, generates a first filter associated with a first memory, and generates one or more second filters each associated with a corresponding memory. Each second filter is generated with a size at least partially accommodated within the storage capacity of the associated memory. The system determines absence of items from a set by selectively probing the generated filters in an ascending order of data access speed to the associated memories and aborts further probing for an item in response to a probe indicating that the item is absent from the set. Embodiments of the present invention further include a method and computer program product for testing set membership in substantially the same manners described above.

Abstract: In one embodiment, a method includes adding, by a computer processor, two or more compressed columns to one or more pages of a database. The adding is performed in parallel by a plurality of page-formatter threads. Each page-formatter thread adds data to the database from no more than a single compressed column.

Abstract: Embodiments relate to the orchestration of data shuffling among memory devices of a non-uniform memory access device. An aspect includes a method of orchestrated shuffling of data in a non-uniform memory access device includes running an application on a plurality of threads executing on a plurality of processing nodes and identifying data to be shuffled among the plurality of processing nodes. The method includes registering the data to be shuffled and generating a plan for orchestrating the shuffling of the data. The method further includes disabling cache coherency of cache memory associated with the processing nodes and shuffling the data among all of the memory devices upon disabling the cache coherency, the shuffling performed based on the plan for orchestrating the shuffling. The method further includes restoring the cache coherency of the cache memory based on completing the shuffling of the data among all of the memory devices.

Abstract: A method of orchestrated shuffling of data in a non-uniform memory access device that includes a plurality of processing nodes includes running an application on a plurality of threads executing on the plurality of processing nodes and identifying data to be shuffled from source threads running on source processing nodes among the processing nodes to target threads executing on target processing nodes among the processing nodes. The method further includes generating a plan for orchestrating the shuffling of the data among the all of the memory devices associated with the threads and shuffling the data among all of the memory devices based on the plan.

Abstract: System, method and computer program products for storing data by computing a plurality of hash functions of data values in a data item, and determining a corresponding memory location for one of the plurality of hash functions of data values in the data item. Each memory location is of a cacheline size wherein a data item is stored in a memory location. Each memory location can store a plurality of data items. A key portion of all data items is contiguously stored within the memory location, and a payload portion is contiguously stored within the memory location. Payload portions are packed as bit-aligned in a fixed-sized memory location, comprising a bucket in a bucketized hash table, each bucket sized to store multiple key portions and payload portions that are packed as bit-aligned in a fixed-sized bucket. Corresponding key portions are stored as compressed keys in said fixed-sized bucket.

Abstract: Providing database support. A first group of data are received, the first group of data being expressed in a first format, and a second group of data are received, the second group of data being expressed in a second format, the second format being different from the first format. The first and second groups of data are merged, and are represented in at least one common column. Such representing includes: maintaining the first and second formats; and providing a tuple map which provides reference to the first and second formats.