Balancing Caching Load In A Peer-To-Peer Based Network File System

• US 20120059864A1
• Filed: 08/29/2011
• Published: 03/08/2012
• Est. Priority Date: 09/03/2010
• Status: Active Grant

• Current Assignee: Toshiba Corporation
• Original Assignee: Marvell World Trade Limited (Marvell Technology Group Limited)
• Inventors: Gole, Abhijeet P., Bandyopadhyay, Somenath, Bhatt, Jayesh B.
• Granted Patent: US 8,463,788 B2
• Time in Patent Office: Days
• US Class Current: 707/827
• CPC Class Codes: H04L 67/1008 based on parameters of servers, e.g. available memory or workload monitoring of computer activity G06F11/30
• CPC Class Codes: H04L 67/1097 for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS]

Abstract

Systems and techniques relating to network file systems for balancing caching load in peer-to-peer based network file systems are described. In one aspect, a method includes maintaining, by a cluster containing two or more computer systems, information about files cached at a network that includes three or more computer systems configured to cache data associated with a file server system. The method also includes receiving, from one of the computer systems of the network, a request to identify at least one computer system of the network that caches a specified file. Further, the method includes identifying, by the cluster in response to the received request, one or more computer systems of the network that cache the specified file based on the maintained information. Furthermore, the method includes providing, by the cluster to the requesting computer system, information referencing at least the identified one or more computer systems of the network.

Claims

1. A system comprising:

a cluster comprising two or more computer systems, the cluster being communicatively coupled with a network that includes three or more computer systems configured to cache data associated with a file server system, each of the two or more computer systems of the cluster is configured to

maintain information about files cached at the three or more computer systems of the network,

receive, from one of the three or more computer systems of the network, a request to identify at least one computer system of the network that caches a specified file,

identify, in response to the received request, one or more computer systems of the network, other than the requesting computer system, that cache the specified file based on the maintained information, and

provide, to the requesting computer system, information referencing at least the identified one or more computer systems of the network.

2. The system of claim 1, where at least one computer system of the cluster is part of the network that includes the three or more of the computer systems configured to cache the data associated with the file server system.

3. The system of claim 2, where all computer systems of the cluster are part of the network that includes the three or more of the computer systems configured to cache the data associated with the file server system.

4. The system of claim 2, where the requesting computer system is not included in the two or more computer systems of the cluster.

5. The system of claim 1, where the provided information referencing at least the identified one or more computer systems of the network includes permission information associated with an instance of the specified file that is cached at the identified one or more computer systems of the network.

6. The system of claim 1, where each of the two or more computer systems of the cluster is further configured to

monitor caching load of each of the three or more computer systems of the network, and

in response to the provided information referencing multiple computer systems of the network that cache the specified file, prioritize the provided information based on respective caching loads of the multiple computer systems referenced by the provided information.

7. The system of claim 1, where each of the two or more computer systems of the cluster is further configured to

receive from the requesting computer system a notification that the specified file was retrieved from a first computer system referenced by the provided information, and

update the maintained information in accordance with the received notification, the update relating to caching load at the requesting computer system and to caching activity at the first computer system.

8. The system of claim 1, where each of the two or more computer systems of the cluster is further configured to

receive from the requesting computer system a notification that the specified file was absent at a first computer system referenced by the provided information and that the specified file was retrieved from a second computer system referenced by the provided information, and

update the maintained information in accordance with the received notification, the update relating to respective caching loads at the requesting computer system and at the first computer system, and to caching activity at the second computer system.

9. The system of claim 1, where the cluster has a single identifier to be used by the three or more computer systems of the network that are not part of the cluster, and each of the two or more computer systems of the cluster is further configured to

respond to the request based on recognition of the single identifier.

10. A device comprising:

memory and circuit electronics, the device being configured to connect with a network that includes a caching load balancer and three or more devices, which include the device, configured to cache data associated with a file server system, such that the devices have respective connections with the file server system, the circuit electronics being configured to

detect that a specified file is not cached on the memory of the device,

request, from the caching load balancer in response to the detection, information that identifies at least one of the three or more devices of the network that cache the specified file,

receive from the caching load balancer a response including information identifying one or more devices of the three or more devices of the network that cache the specified file,

communicate with at least one device of the one or more devices referenced by the response received from the caching load balancer to retrieve the specified file,

retrieve and cache the specified file based on the communication with the at least one device referenced by the response received from the caching load balancer, and

provide the caching load balancer a notification including an identity of the at least one device referenced by the response received from the caching load balancer with which the device has communicated to retrieve the specified file, and a confirmation that the specified file is cached on the memory of the device.

11. The device of claim 10, where the information included in the response received from the caching load balancer includes permission information associated with an instance of the specified file that is cached at the identified one or more devices of the network, and the circuit electronics are further configured to

prior to detecting that the specified file is not cached on the memory of the device, request to access the specified file to perform a given action, and

select the at least one device for communication based on a permission associated with an instance of the specified file cached at the at least one device meeting a permission corresponding to the given action.

12. The device of claim 11, where

the information included in the response received from the caching load balancer identifies multiple devices, and

includes respective caching loads of the identified multiple devices, and the circuit electronics are further configured to carry out the communication with the at least one device of the identified multiple devices in an order based on criteria including

a predefined quantity of network hops between the device and the identified multiple devices, and

a predefined caching load level of the identified multiple devices.

13. The device of claim 10, where

the communication with the at least one device referenced by the response received from the caching load balancer includes

a request transmitted to a first device referenced by the response received from the caching load balancer to provide the specified file, and

a response received from the first device including the specified file, and the notification to the caching load balancer includes the identity of the first device.

14. The device of claim 10, where

the communication with the at least one device referenced by the response received from the caching load balancer includes

a request transmitted to a first device referenced by the response received from the caching load balancer to provide the specified file,

a response received from first device to notify the device that the specified file is missing at the first device, and

a request transmitted, upon receipt of the response from the first device, to a second device referenced by the response received from the caching load balancer to provide the specified file, and

the notification to the caching load balancer includes the identities of the first and second devices and information that the specified file is missing at the first device.

15. The device of claim 14, where the communication with the at least one device referenced by the response received from the caching load balancer further includes a response received from the second device including the specified file.

16. The device of claim 14, where

the information included in the response received from the caching load balancer identifies the first and second devices as being the only devices of the network that cache the specified file,

the communication with the at least one of the devices referenced by the response received from the caching load balancer further includes a response received from the second device to notify the device that the specified file is missing at the second device, and

the notification to the caching load balancer further includes information that the specified file is missing at the second device, and

the circuit electronics are further configured to

in response to having determined that the specified file is missing at the devices referenced by the response received from the caching load balancer, request the specified file from the file server system, and

receive from the file server system a response including the specified file.

17. The device of claim 10, where

the circuit electronics are further configured to

store the information included in the response received from the caching load balancer identifying the one or more devices of the three or more devices of the network that cache the specified file,

detect a cache miss upon an attempt to locally access the specified file, and

determine, responsive to having detected the cache miss, whether the stored information identifying the one or more devices of the three or more devices of the network that cache the specified file is fresh or stale, prior to using the stored information to retrieve the specified file, and

the stored information identifying the one or more devices of the three or more devices of the network that cache the specified file is fresh if the information was stored for a time interval shorter than a predefined duration, and is stale if the information was stored for a time interval longer than or equal to the predefined duration.

18. The device of claim 17, where

responsive to a result of the determination being that the stored information is stale, the circuit electronics are configured to

discard the stored information, and

request, from the caching load balancer, information that identifies at least one of the three or more devices of the network that currently cache the specified file, and

responsive to a result of the determination being that the stored information is fresh, the circuit electronics are configured to communicate with at least one device referenced by the stored information identifying the one or more devices of the three or more devices of the network that cache the specified file to retrieve the specified file.

19. The device of claim 18, where the device communicates with only one device referenced by the stored information, and the circuit electronics are configured to, responsive to the communication conveying that the specified file is missing at the one device,

discard the stored information, and

request, from the caching load balancer, information that identifies at least one of the three or more devices of the network that currently cache the specified file.

20. A method comprising:

maintaining, by a cluster comprising two or more computer systems, information about files cached at a network that includes three or more computer systems configured to cache data associated with a file server system;

receiving, by the cluster from one of the three or more computer systems of the network, a request to identify at least one computer system of the network that caches a specified file;

identifying, by the cluster in response to the received request, one or more computer systems of the network, other than the requesting computer system, that cache the specified file based on the maintained information; and

providing, by the cluster to the requesting computer system, information referencing at least the identified one or more computer systems of the network.