Abstract: Described are improved systems, computer program products, and methods for obtaining space usage information within a clustered database system. Some approaches provide an improved algorithm and structure that gives the ability to compute the latest and accurate space usage with only in-memory operations.

Abstract: Techniques are provided for optimizing storage of database data records in segments using sub-segments. A segment may be comprised of a plurality of extents that contain data blocks and a plurality of references to sub-segments. Sub-segments are containers that contain other extents. A request to insert a set of records into a segment may be received by a database management system. Upon receiving the request, a particular sub-segment is selected to temporarily store the set of records. The set of records are inserted into data blocks belonging to an extent of the sub-segment. Access frequency for the set of records is monitored to determine whether the access frequency is less than a first threshold. When the access frequency of a subset of records is less than a first threshold, the subset of records is compressed and inserted into a particular extent of a particular segment.

Abstract: A shared-nothing database system is provided in which parallelism and workload balancing are increased by assigning the rows of each table to “slices”, and storing multiple copies (“duplicas”) of each slice across the persistent storage of multiple nodes of the shared-nothing database system. When the data for a table is distributed among the nodes of a shared-nothing system in this manner, requests to read data from a particular row of the table may be handled by any node that stores a duplica of the slice to which the row is assigned. For each slice, a single duplica of the slice is designated as the “primary duplica”. All DML operations (e.g. inserts, deletes, updates, etc.) that target a particular row of the table are performed by the node that has the primary duplica of the slice to which the particular row is assigned. The changes made by the DML operations are then propagated from the primary duplica to the other duplicas (“secondary duplicas”) of the same slice.

Abstract: Described are improved systems, computer program products, and methods for obtaining space usage information within a clustered database system. Some approaches provide an improved algorithm and structure that gives the ability to compute the latest and accurate space usage with only in-memory operations.

Abstract: Herein is database query acceleration from dynamic discovery of whether contents of a persistent column can be stored in an accelerated representation in storage-side memory. In an embodiment, based on data type discovery, a storage server detects that column values in a persistent column have a particular data type. Based on storage-side metadata including a frequency of access of the persistent column as an offload input column for offload computation requests on a certain range of memory addresses, the storage server autonomously decides to generate and store, in storage-side memory in the storage server, an accelerated representation of the persistent column that is based on the particular data type. The storage server receives a request to perform an offload computation for the offload input column. Based on the accelerated representation of the persistent column, execution of the offload computation is accelerated.

Abstract: A multiple-tier operation evaluates a query across storage tiers in columnar format. A database server receives from a client a query for reading values from a set of columns of a database table. The multiple-tier operation comprises accessing a first subset of rows for the set of columns in columnar format in a first tier to generate a first set of results and accessing a second subset of rows for the set of columns in columnar format in a second tier to generate a second subset of results. The multiple-tier operation further comprises aggregating the first set of results and the second set of results to form a query result set and returning the query result set to the client.

Abstract: Techniques are provided for optimizing storage of database records in segments using sub-segments. A base segment is a container used for storing records that belong to a database object. A database management system receives a request to load, into the database object, a first set of records that are in a first state. In response to receiving the request, the system generates a new sub-segment, which is a container that is separate from the base segment. The system stores the first set of records, in their first state, within the sub-segment. The system then monitors one or more characteristics of the database system. In response to the one or more characteristics satisfying criteria, the system performs a migration of one or more records of the first set of records from the sub-segment to the base segment. During the migration, the system converts the one or more records from the first state to a second state and stores the one or more records, in their second state, in the base segment.

Abstract: Techniques are provided for implementing an in-memory columnar data store that is configured to either grow or shrink in response to performance prediction data generated from database workload information. A system maintains allocations of volatile memory from a given memory area for a plurality of memory-consuming components in a database system. The system receives for each memory-consuming component, performance prediction data that contains performance predictions for a plurality of memory allocation sizes for the memory-consuming components. The system determines a target memory allocation for an in-memory columnar data store based on the performance predictions. The system determines an incrementally adjusted amount of memory for the in-memory columnar data store and causes the incrementally adjusted amount to be allocated to the in-memory columnar data store.

Abstract: An approach is described to provide a method, a computer program product, and a computer system to implement hole punching. The described approach provides an automated way to free up space without requiring manual intervention by a DBA to manually reorganize database objects to free up space.

Abstract: A shared-nothing database system is provided in which parallelism and workload balancing are increased by assigning the rows of each table to “slices”, and storing multiple copies (“duplicas”) of each slice across the persistent storage of multiple nodes of the shared-nothing database system. When the data for a table is distributed among the nodes of a shared-nothing system in this manner, requests to read data from a particular row of the table may be handled by any node that stores a duplica of the slice to which the row is assigned. For each slice, a single duplica of the slice is designated as the “primary duplica”. All DML operations (e.g. inserts, deletes, updates, etc.) that target a particular row of the table are performed by the node that has the primary duplica of the slice to which the particular row is assigned. The changes made by the DML operations are then propagated from the primary duplica to the other duplicas (“secondary duplicas”) of the same slice.

Abstract: An approach is described to provide a method, a computer program product, and a computer system to implement hole punching. The described approach provides an automated way to free up space without requiring manual intervention by a DBA to manually reorganize database objects to free up space.

Abstract: A shared-nothing database system is provided in which parallelism and workload balancing are increased by assigning the rows of each table to “slices”, and storing multiple copies (“duplicas”) of each slice across the persistent storage of multiple nodes of the shared-nothing database system. When the data for a table is distributed among the nodes of a shared-nothing system in this manner, requests to read data from a particular row of the table may be handled by any node that stores a duplica of the slice to which the row is assigned. For each slice, a single duplica of the slice is designated as the “primary duplica”. All DML operations (e.g. inserts, deletes, updates, etc.) that target a particular row of the table are performed by the node that has the primary duplica of the slice to which the particular row is assigned. The changes made by the DML operations are then propagated from the primary duplica to the other duplicas (“secondary duplicas”) of the same slice.

Abstract: Techniques are described for querying on hybrid formats and storages by a DBMS to leverage faster mirror format (MF) data and completeness of persistent format (PF) data. In an embodiment, the DBMS receives a query that specifies both in-memory and disk-only columns. The DBMS identifies that a scan-based operation is referencing an in-memory column stored within both MF and PF data. The DBMS may initiate performing the operation on the in-memory column on one of the formats of data. While doing so, the DBMS may determine that the actual improvement metric for the operation on the selected format data is not achieving the desired improvement. If so, the DBMS may dynamically switch the performing of the scan-based operation to the other format of the same data.

Abstract: An approach is described to provide a method, a computer program product, and a computer system to implement hole punching. The described approach provides an automated way to free up space without requiring manual intervention by a DBA to manually reorganize database objects to free up space.

Abstract: A shared-nothing database system is provided in which parallelism and workload balancing are increased by assigning the rows of each table to “slices”, and storing multiple copies (“duplicas”) of each slice across the persistent storage of multiple nodes of the shared-nothing database system. When the data for a table is distributed among the nodes of a shared-nothing system in this manner, requests to read data from a particular row of the table may be handled by any node that stores a duplica of the slice to which the row is assigned. For each slice, a single duplica of the slice is designated as the “primary duplica”. All DML operations (e.g. inserts, deletes, updates, etc.) that target a particular row of the table are performed by the node that has the primary duplica of the slice to which the particular row is assigned. The changes made by the DML operations are then propagated from the primary duplica to the other duplicas (“secondary duplicas”) of the same slice.

Abstract: Row locking is performed at the row level of granularity for database data stored in columnar form. Row level locking entails use of a lock vector that is stored in a compression unit in a data block, the compression unit storing rows in columnar-major format. On an as needed basis, the lock vector is expanded to identify more transactions affecting the rows in the compression unit.

Abstract: A computer program product, system, and computer implemented method for automatic, repetitive, and gradual shrink of very large database object structures. The approach disclosed herein generally comprises an iterative best efforts approach to automatically shrink a large database object structure(s). The process may comprise identifying a large object (LOB) structure shrink event and LOB structure or LOB structure segment to be shrunk. And executing a shrink operation on the LOB structure or LOB structure segment without blocking access to other extents in the LOB structure or LOB segment structure. This process allows an extent to be shrunk and deallocated. As a result, the process can operate to continually clean a LOB structure or LOB structure segment to decrease fragmentation within the extents of the LOB structure or LOB segment structure.

Abstract: Techniques are provided for automatic verification of query results obtained from data formatted different from row-major format. Techniques are also provided for automatic triage of query results obtained from data formatted different from row-major format. In an embodiment, the query results may be obtained from querying data formatted in column-major format. In an embodiment, when the query results are determined to be incorrect, a cause of the error in the query results can be automatically identified.

Abstract: Techniques are described herein for hybrid parallelization of in-memory table scans. Work for an in-memory scan is divided into granules based on a degree of parallelism. The granules are assigned to one or more processes. The work for each granule is further parallelized by dividing the work granule into one or more tasks. The tasks are assigned to one or more threads, the number of which can be dynamically adjusted.

Abstract: Techniques are described for querying on hybrid formats and storages by a DBMS to leverage faster mirror format (MF) data and completeness of persistent format (PF) data. In an embodiment, the DBMS receives a query that specifies both in-memory and disk-only columns. The DBMS identifies that a scan-based operation is referencing an in-memory column stored within both MF and PF data. The DBMS may initiate performing the operation on the in-memory column on one of the formats of data. While doing so, the DBMS may determine that the actual improvement metric for the operation on the selected format data is not achieving the desired improvement. If so, the DBMS may dynamically switch the performing of the scan-based operation to the other format of the same data.