Advanced clock synchronization technique

• US 8,150,800 B2
• Filed: 03/28/2007
• Issued: 04/03/2012
• Est. Priority Date: 03/28/2007
• Status: Active Grant

• Current Assignee: NetApp, Inc.
• Original Assignee: NetApp, Inc.
• Inventors: Webman, Erez, Novick, Yoram
• Primary Examiner(s): Ali, Mohammad
• Assistant Examiner(s): Ruiz, Angelica
• Application Number:
  US11/692,659
• Publication Number:
  US 20080243950A1
• Time in Patent Office: 1,833 Days
• Field of Search: None
• US Class Current: 707/609
• CPC Class Codes: G06F 11/14 Error detection or correction of the data by redundancy in operation G06F11/16 takes precedence

Abstract

An advanced clock synchronization technique is adapted for use with a replication service in a data backup and recovery storage environment. The storage environment includes a plurality of source storage systems and target storage systems, wherein the source storage systems are illustratively embodied as source replication nodes. The advanced clock synchronization technique establishes a software-only, loosely-accurate global clock based on a timestamp and an associated error boundary. Notably, the timestamp and its error boundary are used as global clock information to enable synchronization (i.e., ordering of storage requests) among the source replication nodes and/or the target storage systems, thereby ensuring consistent replication of the storage requests on the target storage systems.

Claims

1. A system configured to provide clock synchronization for use with a replication service in a data backup and recovery storage environment, comprising:

a first source replication node of a plurality of source replication nodes (SRNs) configured to couple to one or more target storage systems of a plurality of target storage systems over a network, the plurality of SRNs configured to replicate storage requests on the one or more target storage systems, wherein a first SRN of the plurality of SRNs comprises a first write interception component, a first hardware clock component (HClock) configured to generate a timestamp, and a first clock node component (CNode) in cooperating relation to the first write interception component and the first HClock, the first CNode configured to,

interact with a second CNode of a second SRN of the plurality of SRNs to elect the first CNode as a master CNode,

synchronize a second HClock of the second SRN to the first clock HClock by the second SRN querying a current time from the master CNode,

provide the second CNode with the timestamp from the first HClock in response to the query, and

modify the second HClock according to the timestamp and an error boundary of the second HClock; and

wherein at least one of the CNodes is configured to generate a time identifier (ID) data structure configured to store at least one of the timestamp and the error boundary.

2. The system of claim 1 wherein the storage requests are write operations.

3. The system of claim 2 wherein synchronization orders the write operations.

4. The system of claim 1 wherein each SRN comprises a write interception component configured to intercept the storage requests, each SRN configured to replicate the storage requests to the one or more target storage systems.

5. The system of claim 4 wherein the error boundary is defined by a maximum drift of the timestamp.

6. The system of claim 1 wherein the master CNode comprises a software module configured to interact collectively with each other CNode of the plurality of SRNs to provide an interface to retrieve the current time.

7. The system of claim 1 wherein the time (ID) data structure further comprises a local sequence number (LSN) field containing a counter value that is incremented when the first CNode is queried to return the current time.

8. The system of claim 1 wherein the master node clock component is further configured to set an error field of the time ID structure to zero.

9. The system of claim 1 wherein the time ID data structure further comprises:

a major sequence number (MSN) field containing a counter value that increases in response to one or more low-frequency control events.

10. The system of claim 9 wherein the one or more low-frequency control events comprises a master CNode replacement and a master CNode restart.

11. The system of claim 1 further comprising:

one or more hosts configured to execute applications to issue the storage requests to access data containers exported by the plurality of source storage systems; and

a switching fabric configured to couple the one or more hosts to the plurality of source storage systems.

12. The system of claim 11 wherein a second write interception component resides in one of a host of the one or more hosts, the switching fabric and a source storage system.

13. The system of claim 12 wherein the host is a general-purpose computer and wherein the second write interception component is embodied as a device driver of the general-purpose computer.

14. The system of claim 12 wherein the switching fabric comprises a network switch and wherein the write interception component is embodied as a software module of the network switch.

15. The system of claim 12 wherein the source storage system comprises a virtualization system and wherein the first write interception component is embodied as a software module disposed over the virtualization system.

16. A method for providing clock synchronization for use with a replication service in a data backup and recovery storage environment, comprising:

coupling a first replication node of a plurality of source replication nodes (SRNs) to one or more target storage systems of a plurality of target storage systems over a network;

replicating a storage requests on the one or more target storage systems by the plurality of SRNs, wherein a first SRN of the plurality of SRNs comprises a first write interception component, a first hardware clock component (HClock) configured to generate a timestamp, and a first clock node component (CNode) in a cooperating relation to the first write interception component and the first HClock;

electing the first CNode as a master CNode;

synchronizing a second HClock of a second SRN of the plurality of SRNs to the first HClock by the second SRN querying a current time from the master CNode;

providing a second CNode of the second SRN with the timestamp from the first HClock;

modifying the second HClock according to the timestamp and an error boundary of the second HClock; and

generating, by at least one of the CNodes, a time identifier (ID) data structure configured to store at least one of the timestamp and the error boundary.

17. The method of claim 16 further comprising replicating the storage requests to the one or more target storage systems.

18. The method of claim 16 further comprising defining the error boundary by a maximum drift of the timestamp.

19. The method of claim 16 further comprising interacting by the first CNode collectively with each other CNode of the plurality of SRNs to provide an interface to retrieve the current time.

20. The method of claim 16 further comprising incrementing a counter value of a local sequence number (LSN) field of the time ID data structure when the first CNode is queried to return the current time.

21. The method of claim 16 further comprising

incrementing a counter value of a major sequence number (MSN) field of the time ID data structure in response to one or more low-frequency control events.

22. A computer readable storage media containing executable program instructions for execution by a processor, the computer readable storage media comprising:

program instructions that operatively couple a first source replication node of a plurality of source replication nodes (SRNs) to one or more target storage systems of a plurality of target storage systems over a network, wherein a first SRN of the plurality of SRNs comprises a first write interception component, a first hardware clock (HClock) configured to generate a timestamp, and a first clock node (CNode) in cooperating relation to the first write interception component and the first HClock;

program instructions that elect the first CNode as a master CNode;

program instructions that synchronize a second HClock of a second SRN of the plurality of SRNs to the first HClock by the second SRN querying a current time from the master CNode;

program instructions that provide a second CNode of the second SRN with the timestamp from the first HClock;

program instructions that modify the second HClock according to the timestamp and an error boundary of the second HClock; and

program instructions that generate, by at least one of the CNodes, a time identifier (ID) data structure configured to store at least one of the timestamp and the error boundary.