Sap Se

Hyeong Seog Kim

Yong Sik Kwon

Ji Hoon Jang

Seyeong Bae

Won Wook Hong

Joo Yeon Lee

Jaeyun Noh

Beomsoo Kim

Seungho Yoo

Juchang Lee

Sang Kyun Cha

Kyu Hwan Kim

Taehyung Lee

Jane Jung Lee

Jeong Hee Won

Marcel Birn

Juhyeon Jeong

Younkyoung Lee

Yu Gao

Junpeng Liu

Zhifeng Xu

Disclosed techniques and solutions can provide improved snapshot replication. Typically, an initial replica obtained using snapshot replication is periodically updated. However, the update process can unnecessarily consume computing resources if data in a source data object has not changed with respect to data in a replica data object. Disclosed techniques check to determine whether a snapshot replica is out of date before obtaining a new snapshot. The checks can be performed on manual request or on the occurrence of triggers, such as receiving a query that accesses the replica data object or according to a schedule. Information for current and prior versions of the remote data object can be compared to determine whether a replica is out of date, such as digest values of contents of the remote data object or timestamps associated with the remote data object.

Change-aware snapshot replication

Systems and methods include determination of a first encryption key associated with a data page of a row store database table stored in a volatile memory, based on a header of the data page, encryption of a body of the data page using the first encryption key, and storage of an encrypted data page comprising the header and the encrypted body in a persistent storage system.

Native multi-tenant row table encryption

In some implementations, there is provided a method including receiving a request to provide a local database system with smart data access to a database table stored at a remote database system; executing, by the local database system, a series of one or more fetches, each of which obtains a chunk of the database table stored at the remote database system, such that a corresponding result set for each fetch causes the remote database system to fetch and materialize a corresponding chunk of the database table rather than the database table in its entirety; and reading, by the local database system, a first chunk obtained from the database table stored at the remote database system to form, at least in part, the local copy at the local database system.

Performance optimized pipelined execution for federated databases

Techniques and solutions are provided for providing federated data access to parameterized data objects. At a local system, a virtual parameterized data object is created. A remote computing system is contacted to obtain parameters used by a parameterized data object of the remote computing system to which the virtual parameterized data object corresponds. Parameter information received from the remote system is stored in a definition of the virtual parameterized data object at the local system. When a request for a database operation involving the virtual parameterized data object is received, the parameter information can be used to determine whether the request is correctly formed, or can be used in preparing a request to be sent to the remote system to be performed using the parameterized data object to obtain information specified in the request for a database operation.

Virtual access to parameterized data objects

The present disclosure provides techniques and solutions for executing requests for database operations involving a remote data source in a system that includes an anchor node and one or more non-anchor nodes. A first request for one or more database operations is received, where at least a first database operation includes a data request for a remote data object. It is determined that the first database operation is not an insert, delete, or update operation, and therefore is assignable to the anchor node or one of the non-anchor nodes. The first database operation is assigned to a non-anchor node for execution. In a particular implementation, for a particular set of requests for a database operation, once an insert, delete, or update operation is received for the remote data object, subsequent operations for the remote data object in the set of requests are assigned to the anchor node for execution.

Flexible routing of database operations accessing remote tables

Methods, systems, and computer-readable storage media for receiving a first storage request indicating a first type of data and a first size, identifying a first sub-space based on the first type, the first sub-space including a dense group of dense segments and a sparse group of sparse segments, and determining that the dense group of the first sub-space is absent a dense segment to store data of the first storage request based on the first size, and in response: determining whether the sparse group of the first sub-space includes a sparse segment to store the data of the first storage request based on the first size, and storing the data of the first storage request to the sparse group of the first sub-space in response to determining that the sparse group of the first sub-space includes a sparse segment to store the data of the first storage request.

Memory allocation and deallocation mechanism to reduce fragmentation and enhance defragmentation performance

Methods, systems, and computer-readable storage media for determining that a first transaction results in a first version of data stored in a page in a database system, the first version being stored in a version space and the data being stored in a table space, in response to the first version, providing a version table for the page, the version table including a set of cells each cell corresponding to a record in the page, setting a version table pointer of the page to point to the version table, and setting a version pointer of a first cell to point to the first version in the version space, the first cell corresponding to a record of the data in the page.

Version table scheme for fast version space access

Methods, systems, and computer-readable storage media for online defragmentation of memory in database systems by applying an IX-lock to each table having data stored in a marked page in a set of marked pages, generating a record map including key-value pairs, each being associated with a record location in a marked page, a value of each key-value pair initially set to a first value, iteratively executing the online defragmentation to delete data from marked pages and add the data to non-sparse pages, at least one iteration including applying a try-lock to a record in a marked page, and at iterations of the online defragmentation, updating the record map to change the value of at least one key-value pair from the first value to the second value, the second value representing that data of a marked page has been deleted from the marked page and added to a non-sparse page.

Efficient and non-disruptive online defragmentation with record locking

Restoring a database image of a secondary database system comprises initializing the database image from a recovery image that includes metadata logs and non-metadata logs, of which at least one metadata log includes an undo log of non-committed data that is applied to the initialized database image. A first pagelist and first indexes are then rebuilt based on first data segments of the database image and the corresponding metadata objects. The database image is then updated by replaying the metadata redo log generated from the undo log, the metadata logs, and the non-metadata logs. Then, a second pagelist and second indexes for the database system are rebuilt based on second data segments of the database image and the corresponding non-metadata objects, while each of the first data segments is statically separated from the second data segments in physical memory space. Related apparatus, systems, techniques and articles are also described.

Physio-logical logging for in-memory row-oriented database system

A system, computer-implemented method, and computer-program product embodiments for rebuilding database indexes is disclosed. A pre-loader thread rebuilds a plurality of indexes in the background once a database is initialized. Simultaneously, a query processor receives queries from a user/client of a database. If an index that is required by the query processor has not yet been rebuilt by the pre-loader, a request to rebuild the index is sent to an asynchronous index loader. The asynchronous index loader places the request in a queue, processes the queue based on available system resources, and then rebuilds the requested indexes. Once the necessary indexes have been rebuilt, a wake-up signal is sent to the query processor which then proceeds to execute the query from a user/client.

Asynchronous index loading for database computing system startup latency managment

Improved database consistency and consistency checking may be provided by cross-referenced split data storage for irregular fields using auxiliary pointers. A fixed-size memory slot may be allocated to a database table in a database system, where the fixed-size memory slot has an address. A variable-size memory slot may be allocated to the database table, where the variable-size memory slot has a different address. For an irregular field in the database table, one or more properties of the irregular field may be stored in the fixed-size memory slot. The one or more properties may include a first pointer which may have the address of the variable-size memory slot. For the irregular field in the database table, at least a portion of data of the irregular field and a second pointer, which may have the address of the fixed-size memory slot, may be stored in the variable-size memory slot.

Cross-referenced irregular field storage in databases

A computer implemented method is provided. The method comprises initializing a first database image of a first database system based on a recovery image comprising committed transactions, visible uncommitted transactions, and undo logs associated with the visible uncommitted transactions, undoing the visible uncommitted transactions using the undo logs, generating redo logs from the undo logs, replaying the redo logs to create invisible uncommitted transactions, and after replaying the redo logs, replaying transaction logs received from a second database system. Related apparatus, systems, techniques and articles are also described.

Version space reconstruction during database initialization

Technologies are described for performing proactive memory allocation (e.g., pre-allocation). Proactive memory allocation (e.g., proactive memory page allocation) can be provided for applications, such as database applications. For example, an application can be associated with a free memory pool (e.g., a free memory pool containing free memory pages). A dedicated thread of the application can monitor the free memory pool and perform proactive memory allocation when needed (e.g., when the amount of memory in the pool is low). For example, the dedicated thread can obtain new free memory from the operating system and add it to the pool.

Growing community of inventors

Seoul, South Korea

Hyeong Seog Kim