METHOD AND SYSTEM FOR DATA RECOVERY IN A DATA SYSTEM

• US 20170046234A1
• Filed: 04/03/2015
• Published: 02/16/2017
• Est. Priority Date: 04/03/2015
• Status: Active Grant

• Current Assignee: Yahoo Assets LLC
• Original Assignee: Yahoo! Inc. (Apollo Global Management, Inc.)
• Inventors: Wang, Dongyue, Ge, Sihan, Liang, Wei, Yang, Guangxin, Wang, Xinyu
• Granted Patent: US 10,459,805 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 11/1464 for networked environments
• CPC Class Codes: G06F 11/1474 in transactions G06F16/20 takes precedence
• CPC Class Codes: G06F 16/2365 Ensuring data consistency and integrity
• CPC Class Codes: G06F 2201/84 Using snapshots, i.e. a logical point-in-time copy of the data

Abstract

The present teaching relates to request management and data recovery in a data system. In one example, a request is received for a transaction of data by a first node in the data system. A second node in the data system is determined based on the request. The second node is to carry out the transaction. The request is stored into a first request queue at the first node. The request is sent to the second node. A notice is obtained indicating that information associated with the transaction has been persistently stored. The request is removed from the first request queue in response to the notice.

Claims

1. A method, implemented on a machine having at least one processor, storage, and a communication platform connected to a network for managing a request in a data system comprising a plurality of nodes, the method comprising:

receiving, by a first node in the data system, a request for a transaction of data;

determining a second node in the data system based on the request, wherein the second node is to carry out the transaction;

storing the request into a first request queue at the first node;

sending the request to the second node;

obtaining a notice indicating that information associated with the transaction has been persistently stored; and

removing the request from the first request queue in response to the notice.

2. The method of claim 1, further comprising:

executing the transaction based on the request;

storing the transaction persistently; and

sending the notice to the second node after the transaction is stored persistently, wherein the second node stores the request into a second request queue at the second node and removes the request from the second request queue in response to the notice.

3. The method of claim 1, wherein the second node:

executes the transaction based on the request;

stores the transaction persistently; and

sends the notice to the first node after the transaction is stored persistently.

4. The method of claim 1, further comprising:

performing a snapshot of a memory state of the first node; and

storing the snapshot persistently.

5. The method of claim 1, further comprising assigning an identity to the transaction, wherein the identity distinguishes the transaction from other transactions in the data system and indicates an order for an execution of the transaction relative to executions of other transactions.

6. The method of claim 1, further comprising sending the request to a third node in the data system.

7. A method, implemented on a machine haying at least one processor, storage, and a communication platform connected to a network for data recovery at a first node in a data system comprising a plurality of nodes and a persistent storage, the method comprising:

detecting, at the first node, a failure in connection with first data;

obtaining information associated with a most recent transaction related to the first node from the persistent storage;

requesting each of the other nodes in the data system to transmit one or more transaction requests previously sent to the first node after the most recent transaction;

receiving, from at least one of the other nodes, the one or more transaction requests;

determining a sequence of one or more transactions associated with the one or more transaction requests; and

executing the one or more transactions according to the sequence in order to recover the first data at the first node.

8. The method of claim 7, wherein the obtaining comprises:

restoring a latest snapshot of a memory state of the first node from the persistent storage;

determining a first transaction associated with the latest snapshot;

retrieving, from the persistent storage, one or more second transactions based on the first transaction; and

executing the one or more second transactions, wherein the most recent transaction is a transaction executed last among the one or more second transactions.

9. The method of claim 8, wherein:

the first transaction is a transaction last executed before the latest snapshot was taken;

the one or more second transactions were executed previously in an order following the first transaction; and

the one or more second transactions are executed according to the order.

10. The method of claim 7, further comprising:

obtaining a request for a new transaction; and

holding the request until the first data at the first node is recovered.

11. The method of claim 7, further comprising de-duplicating a transaction request among the one or more transaction requests when multiple copies of the transaction request are received.

12. A system having at least one processor, storage, and a communication platform connected to a network for managing a request in a data system comprising a plurality of nodes, the system comprising:

a request analyzer located in a first node in the data system and configured for receiving a request for a transaction of data;

a coordinate site determiner located in the first node and configured for determining a second node in the data system based on the request and storing the request into a first request queue at the first node, wherein the second node is to carry out the transaction;

a transaction forwarding unit located in the first node and configured for sending the request to the second node;

a transaction obtaining controller located in the first node and configured for obtaining a notice indicating that information associated with the transaction has been persistently stored; and

a request queue clearing unit located in the first node and configured for removing the request from the first request queue in response to the notice.

13. The system of claim 12, wherein at least one of the first node and the second node performs the following:

executing the transaction based on the request;

storing the transaction persistently; and

generating the notice after the transaction is stored persistently.

14. The system of claim 12, further comprising at least one of:

a snapshot performer configured for performing a snapshot of a memory state of the first node and storing the snapshot persistently; and

a transaction identity assigning unit configured for assigning an identity to the transaction, wherein the identity distinguishes the transaction from other transactions in the data system and indicates an order for an execution of the transaction relative to executions of other transactions.

15. A system having at least one processor, storage, and a communication platform connected to a network for data recovery at a first node in a data system comprising a plurality of nodes and a persistent storage, the system comprising:

a recovery initiator located in the first node and configured for detecting a failure in connection with first data;

a commit log retrieving unit located in the first node and configured for obtaining information associated with a most recent transaction related to the first node from the persistent storage;

a buffered request demander located in the first node and configured for requesting each of the other nodes in the data system to transmit one or more transaction requests previously sent to the first node after the most recent transaction;

a buffered request receiver located in the first node and configured for receiving, from at least one of the other nodes, the one or more transaction requests;

a transaction sequence determiner located in the first node and configured for determining a sequence of one or more transactions associated with the one or more transaction requests; and

one or more partition execution engines located in the first node and configured for executing the one or more transactions according to the sequence in order to recover the first data at the first node.

16. The system of claim 15, further comprising a snapshot restoring unit configured for restoring a latest snapshot of a memory state of the first node from the persistent storage and determining a first transaction associated with the latest snapshot, wherein:

the commit log retrieving unit is further configured for retrieving, from the persistent storage, one or more second transactions based on the first transaction; and

the one or more partition execution engines are further configured for executing the one or more second transactions, wherein the most recent transaction is a transaction executed last among the one or more second transactions.

17. The system of claim 16, wherein:

the first transaction is a transaction last executed before the latest snapshot was taken;

the one or more second transactions were executed previously in an order following the first transaction; and

the one or more second transactions are executed according to the order.

18. The system of claim 15, wherein the transaction sequence determiner is further configured for de-duplicating a transaction request among the one or more transaction requests when multiple copies of the transaction request are received.

19. A machine-readable, non-transitory and tangible medium having information recorded thereon for managing a request in a data system comprising a plurality of nodes, wherein the information, when read by the machine, causes the machine to perform the following:

receiving, by a first node in the data system, a request for a transaction of data;

determining a second node in the data system based on the request, wherein the second node is to carry out the transaction;

storing the request into a first request queue at the first node;

sending the request to the second node;

obtaining a notice indicating that information associated with the transaction has been persistently stored; and

removing the request from the first request queue in response to the notice.

20. A machine-readable, non-transitory and tangible medium haying information recorded thereon for data recovery at a first node in a data system comprising a plurality of nodes and a persistent storage, wherein the information, when read by the machine, causes the machine to perform the following:

detecting, at the first node, a failure in connection with first data;

obtaining information associated with a most recent transaction related to the first node from the persistent storage;

requesting each of the other nodes in the data system to transmit one or more transaction requests previously sent to the first node after the most recent transaction;

receiving, from at least one of the other nodes, the one or more transaction requests;

determining a sequence of one or more transactions associated with the one or more transaction requests; and

executing the one or more transactions according to the sequence in order to recover the first data at the first node.