Nuodb, Inc.

Hitachi Data Systems Corporation

Dassault Systemes

Trek Skyward Palmer

Alan Bryant

David B Pinkney

James A Starkey

Kevin S Grimaldi

Nathan W Clark

Techniques are disclosed for disconnected operation in a distributed database system. In an embodiment, the distributed database system implements a disconnected mode of operation allowing isolated regions of database nodes to provisionally commit transactions, with the global requirements of those transactions later satisfied by a transparent healing process after network connectivity is reestablished between the regions. The healing process, in turn, enables the distributed database system to construct a consistent global state of the database that accounts for the transactions provisionally-committed in each isolated region during the disconnected mode. Once the healing process completes, database clients 'see' a healed version of the database that simulates or otherwise closely approximates a state of the database had the distributed database system performed the transactions during a normal, fully-connected, mode of operation.

Disconnected operation within distributed database systems

A distributed database processing system for a database composed of data records organized into tables which processes unique index atoms consistently and concurrently. Each attempt to insert a new key value into such a unique index atom in any given node is routed to a unique index atom chairman for processing. The chairman determines whether the request will be granted. If the request is not granted, the requesting unique index atom continues to try to submit the insert. If the request is granted, the requesting unique index atom modifies the content thereof and broadcasts a replicated given unique index atom all other nodes that contain the replicated unique index atom.

Global uniqueness checking in distributed databases

A distributed data processing system for a database composed of data records organized into tables for responding to a request to split an index in a consistent and concurrent fashion. A request to split an index atom at any given node is processed by a designated 'chairman' for that index atom. The chairman splits its index by maintaining a lower portion thereof and by moving the contents of an upper portion to a newly created index atom as the right sibling. Each other node takes appropriate steps to assure orderly transfer of this information from the chairman consistently and concurrently across the distributed database processing system.

Distributed database management system with dynamically split B-Tree indexes

Distributed database management system with dynamically split B-tree indexes

An 'operate with missing region' feature allows a cluster to continue servicing reads for available regions even when some regions are missing. Upon a given node failure condition, the cluster is placed in an effective read-only mode for all regions. The node failure condition typically is one where there has been a failure of an authoritative region copy and no backup copy is then available. Typically, 'read-only' means no client write or update requests succeed while the cluster is in this state. Those requests should fail with an error message that indicates failure because of the read-only status. Preferably, such requests are re-tried. In this mode, regions are only allowed to perform read operations but the cluster continues to operate with missing regions, and missing regions are entered on a region map. The cluster then automatically recovers returning missing regions, after which it leaves the read-only state.

System and method for enhancing availability of a distributed object storage system during a partial database outage

A redundant array of independent nodes are networked together. Each node executes an instance of an application that provides object-based storage. The nodes are grouped into systems each having multiple nodes. Objects in the systems have access control lists (ACLs) associated therewith. A method of providing a search key to each ACL in any one system comprises: selecting a region for each metadata object pertaining to the objects in the system; and generating, for each ACL associated with an object in the system, an ACL ID (identifier) as a search key which includes the region number of the region selected for the metadata object pertaining to the object, a region map level corresponding to the region selected, and a monotonically increasing counter value that is automatically incremented each time an ACL ID is generated. The ACL ID remains unique during reconfiguration via splitting/merging existing distributed portions of the database.

Cluster-wide unique ID for object access control lists

This disclosure enables a user to incrementally increase or decrease the size of the cluster and, in particular, to incrementally increase or decrease the number of map regions implemented in a metadata manager associated with the cluster while maintaining availability of the cluster. In a representative embodiment, a configurable target map size is identified, either manually (e.g., by the user) or programmatically. If the target map size is found to be within an acceptable value (given the available node configuration), a procedure to create a new map is triggered. The new map preferably is generated as follows. Initially, one or more new temporary region(s) are created either to split or join the existing regions (depending on whether the map is being increased or decreased). For each temporary region, the process also generates MDPL backup regions. Metadata from an original region (or regions) is then migrated to the temporary region(s). Once the migrations have been completed, the old region or regions (from which the metadata is migrated) are removed and what were the temporary regions are, in effect, promoted. The result is the new map.

System and method for managing scalability in a distributed database

Cluster recovery time is reduced by an incremental refresh technique that is described herein. The goal of the technique is to salvage the portion of the redundant distributed database lost (e.g., during a fault) by performing an incremental refresh of only the updates occurring to that portion of the database since the time of loss.

System and method for managing integrity in a distributed database

An 'operate with missing region' feature of this disclosure allows the cluster to continue servicing reads for available regions even when some regions are missing. In particular, upon a given node failure condition, the cluster is placed in an effective read-only mode for all regions. The node failure condition typically is one where there has been a failure of an authoritative region copy and no backup copy is then available. As used herein, 'read-only' means that no client write or update requests will succeed while the cluster is in this state. In particular, those requests should fail with an error message (or the like) that indicates failure because of the read-only status. Preferably, such requests are then re-tried. In this mode, all regions are only allowed to perform read operations. During the read-only state, the cluster continues to operate with missing regions, and missing regions are entered on the region map. The cluster then automatically recovers returning missing region(s), after which is leaves the read-only state.

Growing community of inventors

Cambridge, MA, United States of America

Trek Skyward Palmer