System and method for managing integrity in a distributed database
View Patent ↗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.
1. A non-transitory computer-readable medium having stored thereon instructions that, when executed by a processor, perform a method operative in a computer system having a plurality of nodes networked together in which objects are stored in a set of regions distributed across the computer system, the method comprising:
providing a region map that identifies locations of the set of regions, where the region map contains copies of the regions in one of several states, the states being one of: authoritative, backup, partial and incomplete, and where an incomplete region copy stores a pending update to the region and the partial state is populated from the pending update; and
using the pending update during a restore to repair a region returning to the set of regions by creating a partial (P) copy of the region, updating the partial (P) copy, and converting the partial (P) copy of the region to a backup (B) copy of the region, the partial (P) copy of the region being treated by an authoritative (A) copy of the region as a backup region for all updates;
wherein the partial (P) copy is updated from the pending update in an incomplete (I) copy of the region.
2. The computer-readable medium as described in claim 1 wherein the partial (P) copy is updated by synchronizing a portion thereof from the authoritative (A) copy of the region to undo a last update.
3. The computer-readable medium as described in claim 1 wherein the partial (P) copy of the region is converted to the backup (B) copy of the region when the (P) copy is done updating.
4. The computer-readable medium as described in claim 1 wherein, upon conversion of the partial (P) copy of the region to the backup (B) copy of the region, the incomplete (I) copy is removed.
5. Apparatus for use in a computer system having a plurality of nodes networked together in which objects are stored in a set of regions distributed across the computer system, comprising:
a processor;
computer memory holding computer program instructions that when executed by the processor perform a method, the method comprising:
providing a region map that identifies locations of the set of regions, where the region map contains copies of the regions in one of several states, the states being one of: authoritative, backup, partial and incomplete, and where an incomplete region copy stores a pending update to the region and the partial state is populated from the pending update; and
using the pending update during a restore to repair a region returning to the set of regions by creating a partial (P) copy of the region, updating the partial (P) copy, and converting the partial (P) copy of the region to a backup (B) copy of the region, the partial (P) copy of the region being treated by an authoritative (A) copy of the region as a backup region for all updates;
wherein the partial (P) copy is updated from the pending update in an incomplete (I) copy of the region.
6. The apparatus as described in claim 5 wherein the partial (P) copy is updated by synchronizing a portion thereof from the authoritative (A) copy of the region to undo a last update.
7. The apparatus as described in claim 5 wherein the partial (P) copy of the region is converted to the backup (B) copy of the region when the (P) copy is done updating.
8. The apparatus as described in claim 5 wherein, upon conversion of the partial (P) copy of the region to the backup (B) copy of the region, the incomplete (I) copy is removed.
9. A method operative in a computer system having a plurality of nodes networked together, wherein objects are stored in a set of regions distributed across the computer system, comprising:
providing a region map that identifies locations of the set of regions, where the region map contains copies of the regions in one of several states, the states being one of: authoritative, backup, partial and incomplete, and where an incomplete region copy stores a pending update to the region and the partial state is populated from the pending update; and
using the pending update during a restore to repair a region returning to the set of regions by creating a partial (P) copy of the region, updating the partial (P) copy, and converting the partial (P) copy of the region to a backup (B) copy of the region, the partial (P) copy of the region being treated by an authoritative (A) copy of the region as a backup region for all updates;
wherein the partial (P) copy is updated from the pending update in an incomplete (I) copy of the region.
10. The method as described in claim 9 wherein the partial (P) copy of the region is converted to the backup (B) copy of the region when the (P) copy is done updating.
11. The method as described in claim 9 wherein, upon conversion of the partial (B) copy of the region to the backup (B) copy of the region, the incomplete (I) copy is removed.