Abstract: The subject technology generates a dataset based at least in part on a set of files. The subject technology generates, utilizing a machine learning model, a set of labels corresponding to the dataset. The subject technology filters the dataset using a set of conditions to generate at least a subset of the dataset. The subject technology generates a virtual object based at least in part on the subset of the dataset and the set of labels, where the virtual object corresponds to a selection of data from the dataset. The subject technology trains a second machine learning model using the virtual object and at least the subset of the dataset, where training the second machine learning model includes utilizing streaming file input/output (I/O), the streaming file I/O providing access to at least the subset of the dataset during training.

Abstract: The present invention relates to a system and methodology to facilitate security for data items residing within (or associated with) a hierarchical database or storage structure. A database security system is provided having a hierarchical data structure associated with one or more data items. The system includes a security component that applies a security policy to the data items from a global location or region associated with a database. Various components and processes are employed to enable explicit and/or inherited security properties to be received by and propagated to the data items depending on the type of data structure encountered or processed.

Abstract: In a database system that supports user-defined types, one or more fields of a user-defined type may be designated as containing data that is to be stored as a file outside of a database store, separate from the other fields of the type which are stored within the database. Instead of being stored within the database, the data in such a designated field of an instance of the user-defined type is stored as a file within the file system of a computer. Applications are provided “out of band” access, via the file system of the computer, to the file in which the data of such a field is stored outside the database store. A component of the system intercepts file system open requests and checks security against the database store before allowing the open request to proceed. The component also notifies the database when file system close requests are received.

Abstract: A system and method for expediting the replication of at least one specified object to a replica in a distributed computer system. A source object of a source replica determines that it has an urgent change to propagate through the distributed system, and informs a replication facility at the source replica of the urgent change. The facility extracts, or is provided with, the change information from that object, and the source replica communicates information representative of the change to a destination replica. A replication facility at the destination replica provides the change information to a destination replica object, which uses the information to make itself consistent with the source replica object.

Abstract: An item inheritance system and method are provided. The item inheritance system can be employed to propagate access control information (e.g., an access control list) to one or more item(s), thus facilitating security of item(s). At least one of the item(s) is a compound item. The item inheritance system includes an input component that receives information associated with one or more items. The items can include container(s), object(s) and/or compound item(s). The system can be triggered by a change in security policy to the item(s), for example, adding and/or deleting a user's access to the item(s). Additionally, moving and/or copying a collection of items can further trigger the system. The system further includes a propagation component that propagates access control information to the item(s). For example, the propagation component can enforce the ACL propagation policies when a change to the security descriptor takes place at the root of a hierarchy.

Abstract: Mechanisms and techniques for database fragment cloning and management are provided. A database object, such as a table, rowset or index, is divided into fragments. Each fragment is cloned to create cloned fragments, which operationally are substantially identical to one another. One or more of the cloned fragments may be designated as a primary cloned fragment for performing database operations or as a secondary cloned fragment for serving as backup. Updates to each fragment are implemented on the primary cloned fragment and are then propagated from the primary cloned fragment to the corresponding secondary cloned fragments. A cloned fragment can go offline, becoming unavailable to be updated. When the cloned fragment returns online, the cloned fragment is refreshed with data included in the primary cloned fragment. While being refreshed, the cloned fragment may continue to be updated.

Abstract: Described are sequence numbers for client-server communication, to control a client's use of server resources. A server grants the client credits, and the client consumes a credit for sending each command to the server. Each credit corresponds to a sequence number, with the set of sequence numbers forming a valid command window. The server enforces that for each received command, the command includes a sequence number that is within the valid command window and that the sequence number has not been used with another command. The server may also maintain a maximum window size, such that clients with credits cannot send a command with a sequence number that beyond a maximum sequence number. When incorporated into a data communication protocol, quality of service, combating denial of service, detection of message loss, division of server resources, secure message signing, and other numerous benefits result.

Abstract: Methods and computer-readable media for maintaining transactional link-level consistency between a database and a file system. A file system change is logged in a record of a database log and a file corresponding to the file system change is created in a file system folder. During a restart recovery process, an analysis operation and a conditional redo operation are performed based on the database log, and a conditional redo operation and an undo operation are performed based on the files in the file system folder. An undo operation is then performed based on the database log.

Abstract: A system and method for expediting the replication of at least one specified object to a replica in a distributed computer system. A source object of a source replica determines that it has an urgent change to propagate through the distributed system, and informs a replication facility at the source replica of the urgent change. The facility extracts, or is provided with, the change information from that object, and the source replica communicates information representative of the change to a destination replica. A replication facility at the destination replica provides the change information to a destination replica object, which uses the information to make itself consistent with the source replica object.

Abstract: A system and method for synchronizing a database having metadata and raw data. Metadata may be fully synchronized between two systems, with copies stored on both systems. Raw data may be selectively synchronized, so that a client system need not store all of the raw data locally. When disconnected, the client system may identify certain portions of the raw data to store locally, and such data may be synchronized when reconnected to a server system.

Abstract: A system and method for expediting the replication of at least one specified object to a replica in a distributed computer system. A source object of a source replica determines that it has an urgent change to propagate through the distributed system, and informs a replication facility at the source replica of the urgent change. The facility extracts, or is provided with, the change information from that object, and the source replica communicates information representative of the change to a destination replica. A replication facility at the destination replica provides the change information to a destination replica object, which uses the information to make itself consistent with the source replica object.

Abstract: A file system that does not require unique item names, or any item name at all, is described herein. If an item has an ambiguous name, the file system performs a disambiguating procedure to provide the client (user or application) a conceptually unique name, including a fully qualified path. The file system provides usability features such that the file system maintains compatibility with legacy applications and systems, including creating a synthetic item name when the item has no name, and disambiguating two items having the same name by using a disambiguating character, such as a small integer or alphanumeric character.

Abstract: A file system that does not conflate item lifetime with the item's inclusion in an organizational structure of the file system is described herein. In addition, the file system's organizational structure is not limited to directory trees, but instead may use any directed acyclic graph (DAG). Items, once stored into the file system, may be maintained until affirmatively deleted by a client of the file system, regardless of whether the item is part of a DAG or not. Items may be placed in file regions, which are conceptual workspaces for a client to control item lifetime and to organize items into any DAG structure of the user's choosing. Items may be concurrently stored in multiple DAGs, and each file region may have one or more independent DAGs. An item's placement in a DAG may also be used to govern file properties such as namespace, security, privacy, and read/write attributes.

Abstract: A pipeline I/O mechanism in a network redirector that breaks up large files into sections, and sends write (or read) requests for each section in a pipeline fashion to a network file server, without waiting for a response for a previously-sent section. The pipeline I/O mechanism tracks the returned status of each section for which a request was made, so that success or failure of the request is determined as a whole. The pipeline I/O system and method dramatically increase the performance of network remote I/O operations, particularly with file write operations directed to a file on a network file server.

Abstract: Systems and methodologies for integration of an anti virus AV Plug In(s) as part of an Item Store. The semantics for operation of the AV Plug In(s) are provided by the relational Item Store, via employing a metadata component and a scanning component associated with the Item Store. The metadata component can supply a signature value being associated with the Item Store that can represent the time of scanning of data, and outcome for every scanned item. The scanning component can provide for a queuing of items in the data store in a synchronous and/or asynchronous mode for both scanning and cleaning by the AV Plug In supplied by vendors.

Abstract: A type of an object that can be persisted in a database store is defined by a user. The type definition comprises fields and behaviors, each field having a respective data type. One or more fields of the type definition may be designated as containing data of a type that is to be stored as a file outside of the database store and separately from the other fields of the type definition. When a request is received to store an object that is an instance of the user defined type, the data in any so designated field of the object is stored in a file outside of the database store, preferably within a file system of the computer on which the database store is implemented. The data in each of the other fields of the object is stored within the database store in the usual manner. The database store maintains a link, or reference, between the persisted object and the data of the field that is stored as a file outside of the database store.

Abstract: The sharing model for file system operations is unified with the transaction and locking model of query language statements to provide an overall framework for locking and isolation in a storage platform. Additionally, transactional support is provided for file system operations so that they may be executed in the context of a transaction. Accordingly, a single transaction may include a single file system statement, a single query language statement, multiple file system statements, multiple query language statements, and a combination of file system and query language statements. Furthermore, support is provided for non-transacted file systems statements so that file system statements need not necessarily be executed in the context of a transaction.

Abstract: A type of an object that can be persisted in a database store is defined by a user. The type definition comprises fields and behaviors, each field having a respective data type. One or more fields of the type definition may be designated as containing data of a type that is to be stored as a file outside of the database store and separately from the other fields of the type definition. When a request is received to store an object that is an instance of the user defined type, the data in any so designated field of the object is stored in a file outside of the database store, preferably within a file system of the computer on which the database store is implemented. The data in each of the other fields of the object is stored within the database store in the usual manner. The database store maintains a link, or reference, between the persisted object and the data of the field that is stored as a file outside of the database store.

Abstract: A method and system for providing state-full access to files and resuming access should a connection be broken. A resume key is returned to the client that allows the client to request a duplicate handle to an open file. The duplicate handle can be used to access the file in the same manner as the handle used to open the file. When a connection is broken, the file remains open on the server for a period of time and the state information associated with the file is maintained. Even if a connection is not broken, a client may request one or more duplicate handles and establish other channels (also known as connections) with which to access the file.

Abstract: A system to perform transacted remote file operations over a network includes a client and a server. The client and server each include a transaction manager (TM) and a file system (FS). The client also includes a redirector (RDR), while the server includes a server application (SRV). The RDR receives a request for a remote transacted file operation. In response, the RDR retrieves the transaction from the request. The RDR may use the TM to marshall the transaction for transmission to the server. The RDR sends the transaction to the server over the network. The SRV component receives the transaction, which the TM and FS of the server then use to perform the file operation. The server then returns the file operation result to the client via the network.