Method and system for implementing a redo repeater

Patent metadata

Field	Value
Publication number	US-10102266-B2
Application number	US-201213454799-A
Country	US
Kind code	B2
Filing date	Apr 24, 2012
Priority date	Apr 24, 2012
Publication date	Oct 16, 2018
Grant date	Oct 16, 2018

Abstract

Disclosed are methods and apparatuses to provide a redo repeater that allows for no data loss protection without the performance impact to the primary database even when a significant geographical distance separates the primary and standby databases. The Repeater is a lightweight entity that receives redo from the primary database with the purpose of redistributing that redo throughout the primary/standby system configuration. The Repeater able to extend no data loss protection and switchover functionality to terminal standby databases even though the primary database does not need to have a direct connection with those destinations.

First claim

What is claimed is:

1. A method implemented with a one or more processors, comprising:
maintaining, by the one or more processors, a primary database system, a standby database system, and two or more relay devices, wherein the primary database system includes one or more computer systems that are connected to the standby database system via a network;
the primary database system being located at a first geographic location, the primary database system creating redo records to be replicated at the standby database system using a selected relay device of the two or more relay devices, the primary database system being electrically coupled over a first portion of the network to the two or more relay devices,
the two or more relay devices being located at a second set of geographic locations, the two or more relay devices being electrically coupled over one or more first portions of the network to the primary database system and being electrically coupled over one or more second portions of the network to the standby database system, wherein relay devices of the two or more relay devices forward respective redo records from respective primary database systems to respective standby database systems, and the relay devices of the two or more relay devices do not apply the respective redo records to respective databases at respective relay device, the two or more relay devices comprising at least a first relay device and a second relay device in different geographic locations of the second set of geographic locations, the different geographic locations having different distances from the primary database system corresponding to different latencies between the first relay device and the primary database system and between the second relay device and the primary database system, the first relay device and the second relay device not relying on any power sources shared with the primary database system, wherein the second set of geographic locations are geographically separated from the primary database system, and
the standby database system being located at a third geographic location, the standby database system storing received redo records into a standby database, the standby database system being electrically coupled over at least some of the second portions of the network to the two or more relay devices;

generating, by the one or more processors, redo records at the primary database system for a primary database, the generated redo records corresponding to changes to database data on the primary database to be written to a standby database at the standby database system;
selecting, by the one or more processors, a relay device that does not rely on any power sources shared with the primary database system from at least the first relay device and the second relay device for the primary database of the primary database system, the relay device being selected based at least in part upon both a first network latency between the primary database system and the first relay device and a second network latency between the primary database system and the second relay device, wherein corresponding redo records are committed at the primary database system only after receiving acknowledgement of storage of the corresponding redo records at the selected relay device that does not rely on any power sources;
relaying, by the one or more processors, redo records of the generated redo records for the primary database of the primary database system to the standby database of the standby database system at least by:
receiving, by the one or more processors, the relayed redo records at the relay device;
storing, by the one or more processors, the relayed redo records in persistent storage of the relay device, the persistent storage comprising a non-volatile medium;
sending, by the one or more processors, one or more acknowledgement messages from the relay device to the primary database after storing the relayed redo records in the persistent storage of the relay device, the one or more acknowledgement messages confirming storage of the relayed redo records in the persistent storage of the relay device; and
forwarding, by the one or more processors, the relayed redo records from the relay device to the standby database system, wherein the relay device does not share a common point of failure with either the primary database system or the standby database system;

determining, by the one or more processors, that a new relay device is to replace the selected relay device, wherein the determining whether the new relay device is to replace the selected relay device based in part or in whole upon a detection of a failure of the relay device; and
selecting, by the one or more processors, the new relay device from the two or more relay devices, wherein the new relay device that does not rely on any power sources shared with the primary database system, the relay device being selected based at least in part upon both a network latency between the primary database system and the new relay device, wherein corresponding redo records are committed at the primary database system only after receiving acknowledgement of storage of the corresponding redo records at the new relay device.

2. The method of claim 1 , further comprising:
determining that a new primary database system is to replace the primary database system, wherein determining whether the new primary database system is to replace the primary database system based in part or in whole upon one or more changes involving the primary database;
designating at least one standby database system of a plurality of standby database systems as the new primary database system that replaces the primary database system;
determining whether the relay device services the new primary database system based in part or in whole upon respective geographic distances and respective latencies between the new primary database system and the two or more relay devices; and
when the relay device is determined not to service the new primary database system, selecting a new relay device for the new primary database system based in part or in whole upon the respective geographic distances and the respective latencies between the new primary database system and the two or more relay devices.

3. The method of claim 1 , wherein the relay device does not include a copy of user data tables.

4. The method of claim 1 , wherein the relay device is geographically located at a distance corresponding to a network latency that is less than a local disk latency at the primary database system.

5. The method of claim 1 , wherein the relay device maintains control information corresponding to location, content, and sequence information of associated redo logs.

6. The method of claim 1 , further comprising forwarding redo records from the relay device to a second standby database system, wherein the relay device does not share a common point of failure with the second standby database system.

7. The method of claim 1 , wherein redo records are sent from the primary database system to a relay device synchronously, and are sent from the relay device to the standby database system asynchronously.

8. A computer program product embodied on a non-transitory computer usable medium, the non-transitory computer usable medium having stored thereon a sequence of instructions which, when executed by one or more processors causes the one or more processors to execute a method, the method comprising:
maintaining a primary database system, a standby database system, and two or more relay devices, wherein the primary database system includes one or more computer systems that are connected to the standby database system via a network;
the primary database system being located at a first geogra

Assignees

• Mcgee Steven
• Luo Jiangbin
• Garin Benedicto
• Girkar Mahesh
• Harvey Michael
• Karkhanis Nitin
• Lee Steve
• Loaiza Juan
• Mcguirk Robert
• Yang Hongjie
• Oracle Int Corp

Inventors

• Mcgee Steven
• Luo Jiangbin
• Garin Benedicto
• Girkar Mahesh
• Harvey Michael
• Karkhanis Nitin
• Lee Steve
• Loaiza Juan
• Mcguirk Robert
• Yang Hongjie

Classifications

• G06F17/30581 Primary

  Physics · mapped topic

• G06F16/275 Primary

  Synchronous replication · CPC title