Recovery point objective enforcement

• US 9,805,104 B2
• Filed: 02/15/2013
• Issued: 10/31/2017
• Est. Priority Date: 02/15/2013
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Talius, Tomas, Nosov, Alexander, Narlapuram, Satyanarayana, Dhamankar, Robin D.
• Primary Examiner(s): Burke, Jeffrey A
• Assistant Examiner(s): VU, THONG H
• Application Number:
  US13/768,539
• Publication Number:
  US 20140236891A1
• Time in Patent Office: 1,719 Days
• Field of Search: 707100, 707613, 707610, 707639, 707762, 711162
• CPC Class Codes: G06F 11/00 Error detection; Error correction; Monitoring error detection, correction or monitoring in information storage based on relative movement between record carrier and transducer G11B20/18; monitoring, i.e. supervising the progress of recording or reproducing G11B27/36; in static stores G11C29/00
• CPC Class Codes: G06F 11/202 where processing functionality is redundant redundant communication control functionality G06F11/2005, redundant storage control functionality G06F11/2089
• CPC Class Codes: G06F 11/2038 with a single idle spare processing component
• CPC Class Codes: G06F 11/2048 where the redundant components share neither address space nor persistent storage
• CPC Class Codes: G06F 11/2097 maintaining the standby controller/processing unit updated initialisation or re-synchronisation thereof G06F11/1658 and subgroups
• CPC Class Codes: G06F 11/3006 where the computing system is distributed, e.g. networked systems, clusters, multiprocessor systems multiprogramming arrangements G06F9/46; allocation of resources G06F9/50
• CPC Class Codes: G06F 11/3055 Monitoring arrangements for monitoring the status of the computing system or of the computing system component, e.g. monitoring if the computing system is on, off, available, not available error or fault processing without redundancy G06F11/0703; error detection or correction by redundancy in data representation G06F11/08; error detection or correction by redundancy in operation G06F11/14; error detection or correction by redundancy in hardware G06F11/16
• CPC Class Codes: G06F 16/27 Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
• CPC Class Codes: G06F 16/273 Asynchronous replication or reconciliation

Abstract

A maximum lag between data stores can be specified that corresponds to a recovery point objective defined in a service level agreement. Lag can be monitored during a data replication between a primary data store and a secondary data store located in geographically different regions. Activity on the primary data store including incoming data transactions can be throttled as a function of the lag and the maximum lag.

Claims

1. A method of coordinating transactions committed at a primary data store with a secondary data store, the method involving a device having a processor and comprising:

executing, by the processor, instructions that cause the device to;

store a service level agreement identifying a data loss tolerance of data stored by the primary data store and not the secondary data store;

for respective transactions committed by the primary data store, measure a synchronization lag as a duration between a transaction commit of committed data of the transaction to the primary data store, and synchronization of the committed data stored by the primary data store with the secondary data store;

based upon the synchronization lag between the transaction commit of the committed data and synchronizing the committed data to the secondary data store, estimate a current data loss potential; and

on condition of the current data loss potential exceeding the data loss tolerance specified by the service level agreement, throttle a transaction count of unsynchronized transactions pending at the primary data store until the current data loss potential no longer exceeds the data loss tolerance.

2. The method of claim 1, wherein throttling the transactions on the primary data store further comprises:

responsive to a request to commit a transaction at the primary data store, abort the transaction rather than committing the transaction.

3. The method of claim 1, wherein throttling the transactions on the primary data store further comprises:

reducing a speed at which incoming transactions are committed at the primary data store.

4. The method of claim 3, wherein reducing the speed at which incoming transactions are committed at the primary data store further comprises:

comparing the current data loss potential to the data loss tolerance to determine a proximity of the synchronization lag to a synchronization lag maximum beyond which the current data loss potential exceeds the data loss tolerance; and

reducing the speed at which incoming transactions are committed proportionally with the proximity of the synchronization lag to the synchronization lag maximum.

5. The method of claim 3, wherein reducing the speed at which incoming transactions are committed at the primary data store further comprises:

comparing the current data loss potential to the data loss tolerance to determine a lag rate at which the synchronization lag is approaching a synchronization lag maximum beyond which the current data loss potential exceeds the data loss tolerance; and

reducing the speed at which incoming transactions are committed proportionally with the lag rate at which the synchronization lag is approaching the synchronization lag maximum.

6. The method of claim 1, wherein throttling the transactions on the primary data store further comprises:

reducing a transaction initiation rate at which incoming transactions are initiated at the primary data store.

7. The method of claim 1, wherein:

the secondary data store transmits an acknowledgment of completing synchronizing of the committed data for respective transactions; and

measuring the synchronization lag further comprises;

measuring the synchronization lag as an age of a longest-pending transaction commit for which the secondary data store has not transmitted an acknowledgment of completing synchronization of the committed data for the longest-pending transaction commit.

8. The method of claim 1, wherein throttling the transactions on the primary data store further comprises:

rejecting, rather than accepting, incoming transactions received by the primary data store.

9. The method of claim 1, wherein throttling the transactions on the primary data store further comprises:

reducing a speed at which pending transactions are processed at the primary data store.

10. The method of claim 1, wherein:

synchronizing the primary data store with the secondary data store further comprises;

synchronizing the committed data stored by the primary data store with the secondary data store over a network, wherein synchronizing over the network consumes network capacity at a network bandwidth; and

executing the instructions further causes the device to allocate network bandwidth proportionally to the data loss tolerance of data stored by the primary data store and not the secondary data store.

11. The method of claim 1, wherein reducing the speed at which incoming transactions are committed at the primary data store further comprises:

on condition of the current data loss potential exceeding the data loss tolerance, incrementally reducing the speed at which incoming transactions are committed at the primary data store.

12. The method of claim 1, wherein:

synchronizing the primary data store with the secondary data store further comprises;

synchronizing the committed data stored by the primary data store with the secondary data store over a network, wherein synchronizing over the network consumes network capacity at a network bandwidth; and

executing the instructions further causes the device to, responsive to determining that the synchronization lag is approaching a synchronization lag maximum beyond which the current data loss potential exceeds the data loss tolerance, increase a network bandwidth allocation between the primary data store and the secondary data store.

13. The method of claim 1, wherein:

the transactions are committed by the primary data store on behalf of an application exhibiting a data loss sensitivity; and

estimating the current data loss potential further comprises;

estimating the current data loss potential according to the data loss sensitivity of the application.

14. The method of claim 1, wherein estimating the data current data loss potential further comprises:

estimating a data acquisition rate of the primary data store resulting from committing the transactions; and

estimating the current data loss potential according to the lag and the data acquisition rate of the primary data store.

15. The method of claim 14, wherein:

the transactions are committed by the primary data store on behalf of an application; and

estimating the data acquisition rate further comprises;

estimating the data acquisition rate of the application based upon a chronology factor selected from a chronology factor set comprising;

a current time of year;

a current date;

a current day of week; and

a current time of day.

16. The method of claim 1, wherein:

the transactions are committed by the primary data store on behalf of an application; and

executing the instructions further causes the device to, responsive to detecting a failure of the primary data store, initiate a failover of the application from the primary data store to the secondary data store.

17. A system for coordinating transactions committed at a primary data store with a secondary data store, the system comprising:

a service level agreement that identifies a data loss tolerance of data stored by the primary data store and not the secondary data store;

a transaction monitor that;

for respective transactions committed by the primary data store, measure a synchronization lag as a duration between a transaction commit of committed data of the transaction to the primary data store, and synchronization of the committed data stored by the primary data store with the secondary data store; and

based upon the synchronization lag between the transaction commit of the committed data and synchronizing the committed data to the secondary data store, estimate a current data loss potential; and

a transaction throttle that;

performs a comparison of the current data loss potential to the data loss tolerance;

chooses a transaction throttling rate based on the comparison; and

throttles a transaction count of transactions pending at the primary data source according to the transaction throttling rate.

18. The system of claim 17, wherein choosing the transaction throttling rate based on the comparison further comprises:

determining that the current data loss potential is approaching and has not yet exceeded the data loss tolerance; and

choosing a positive transaction throttling rate, to apply to the transactions at the primary data source while the current data loss potential has not yet exceeded the data loss tolerance, to prevent the current data loss potential from exceeding the data loss tolerance.

19. The system of claim 18, wherein choosing the positive transaction throttling rate, while the current data loss potential has not yet exceeded the data loss tolerance, further comprises:

choosing the positive transaction throttling rate proportional to a proximity of the current data loss potential to the data loss tolerance.

20. A computer-readable storage medium having instructions stored thereon that enable at least one processor of a device to coordinate transactions committed at a primary data store with a secondary data store, by:

storing a service level agreement identifying a data loss tolerance of data stored by the primary data store and not the secondary data store;

for respective transactions committed by the primary data store, measuring a synchronization lag as a duration between a transaction commit of committed data of the transaction to the primary data source, and synchronization of the committed data stored by the primary data store with the secondary data source;

based upon the synchronization lag between the transaction commit of the committed data and synchronizing the committed data to the secondary data store, estimate a current data loss potential of data; and

on condition of the current data loss potential exceeding the data loss tolerance specified by the service level agreement, throttling a transaction count of transactions at the primary data source pending at the primary data store.