LOAD-DERIVED PROBABILITY-BASED DOMAIN NAME SERVICE IN A NETWORK STORAGE CLUSTER

• US 20100023621A1
• Filed: 07/24/2008
• Published: 01/28/2010
• Est. Priority Date: 07/24/2008
• Status: Active Grant

• Current Assignee: NetApp, Inc.
• Original Assignee: NetApp, Inc.
• Inventors: Ezolt, Phillip G., Semke, Jeffrey E.
• Granted Patent: US 8,271,652 B2
• Time in Patent Office: Days
• US Class Current: 709/226
• CPC Class Codes: H04L 61/4511 using domain name system [DNS]
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers
• 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 67/101 based on network conditions
• CPC Class Codes: H04L 67/1019 Random or heuristic server selection
• CPC Class Codes: H04L 67/1023 based on a hash applied to IP addresses or costs
• 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

DNS name resolution is integrated into each node in a network storage cluster, to allow load balancing of network addresses, using a weighted random distribution to resolve DNS requests. A node in the cluster gathers statistics on utilization of resources, such as CPU utilization and throughput, on nodes in the cluster and distributes those statistics to all other nodes. Each node uses the same algorithm to generate weights for the various IP addresses of the cluster, based on the statistics distributed to it. The weights are used to generate a weighted list of available network addresses. In response to a DNS request, a DNS in a given node randomly indexes into the weighted address list to resolve requests to a network address. The weights are chosen so that the DNS is likely to pick an IP address which has a low load, to balance port and node usage over time.

Claims

1. A method comprising:

implementing a domain name service (DNS) within a network storage cluster that includes a plurality of storage server nodes; and

load balancing between the storage server nodes in the cluster, by using a weighted random distribution to resolve client name service requests to network addresses in the storage cluster.

2. A method as recited in claim 1, wherein said load balancing comprises using the same weighted random distribution in each of the plurality of storage server nodes to resolve client name service requests.

3. A method as recited in claim 1, wherein using a weighted random distribution to resolve client name service requests comprises randomly indexing into a weighted list of available network addresses.

4. A method as recited in claim 1, wherein the DNS is implemented in two or more storage server nodes in the network storage cluster.

5. A method as recited in claim 4, wherein the DNS is implemented in each of the storage server nodes in the network storage cluster.

6. A method as recited in claim 5, wherein the weighted random distribution is implemented identically in each of the storage server nodes in the network storage cluster.

7. A method as recited in claim 4, wherein the weighted random distribution is implemented identically in each of the two or more storage server nodes in the network storage cluster.

8. A method as recited in claim 1, further comprising resolving client name service requests to load balance between different ports on a particular storage server node in the network storage cluster.

9. A method as recited in claim 1, wherein the DNS returns a single network address in response to each client name service request.

10. A method as recited in claim 1, further comprising:

gathering statistics at a storage server node in the network storage cluster, relating to throughput and CPU utilization, for each of a plurality of network addresses in the cluster; and

distributing the statistics from said storage server node to each of the other storage server nodes in the cluster.

11. A method as recited in claim 10, further comprising:

determining in each of the storage server nodes a set of weights for the plurality of network addresses, based on the distributed statistics, wherein each of the storage server nodes determines the same set of weights for the plurality of network addresses.

12. A method as recited in claim 11, wherein determining the set of weights comprises, for each of the plurality of network addresses:

computing a first value corresponding to a percentage of CPU capacity being used;

computing a second value corresponding to a percentage of throughput capacity being used; and

computing a third value as a function of the first value and the second value.

13. A method as recited in claim 1, further comprising:

detecting that a network address has been disabled by an administrator; and

removing a network address from a list of candidate network addresses in response to detecting that the network address has been disabled by an administrator.

14. A method comprising:

gathering statistics, at a storage server node in a network storage cluster that includes a plurality of storage server nodes, the statistics relating to CPU utilization and throughput for each of a set of ports in the network storage cluster;

distributing the statistics from said storage server node to each of the other storage server nodes in the cluster;

in each of the storage server nodes in the cluster, determining a set of weights for a set of network addresses of the cluster, based on the distributed statistics, wherein each of the storage server nodes determines the same set of weights for the set of network addresses;

in each of the storage server nodes in the cluster, creating a weighted list of available network addresses, based on the set of weights;

receiving a client name service request at one of the storage server nodes; and

in response to the client name service request, randomly indexing into the weighted list in said one of the storage server nodes, to resolve the client name service request to a network address in the storage cluster.

15. A method as recited in claim 14, further comprising resolving the client name service request to load balance between different ports on a particular storage server node in the network storage cluster.

16. A method as recited in claim 14, further comprising:

detecting that a network address has been disabled by an administrator; and

removing a network address from a weighted list of available network addresses in response to detecting that the network address has been disabled by an administrator.

17. A method as recited in claim 14, wherein the statistics comprise, for each of a plurality of network addresses of the storage cluster:

a first statistic corresponding to a percentage of CPU capacity being used; and

a second statistic corresponding to a percentage of throughput capacity being used.

18. A system comprising:

a plurality of storage server nodes coupled to each other to form a network storage cluster, and

a domain name server (DNS) implemented by each storage server node in the cluster, wherein each of the storage server nodes is capable of resolving a client mount request containing a host name to a network address in the cluster, by using a weighted random distribution.

19. A system as recited in claim 18, wherein using a weighted random distribution comprises randomly indexing into a weighted list of network addresses.

20. A system as recited in claim 18, wherein each of the storage server nodes uses the same weighted random distribution to resolve client name service requests.

21. A system as recited in claim 18, wherein each of the storage server nodes uses an identical copy of said weighted random distribution.

22. A method as recited in claim 18, wherein the DNS returns a single network address in response to each client name service request.

23. A system as recited in claim 18, wherein each of the storage server nodes further comprises:

a gatherer to gather statistics relating to CPU utilization and throughput for each of a plurality of ports in the network storage cluster and to distribute the statistics to another storage server node in the cluster; and

a virtual interface manager to determine a set of weights for a set of network addresses of the cluster, based on the distributed statistics and to provide the weights as input for forming a weighted list of network addresses.

24. A system as recited in claim 23, wherein the statistics comprise, for each of a plurality of network addresses of the storage cluster:

a first statistic corresponding to a percentage of CPU capacity being used; and

a second statistic corresponding to a percentage of throughput capacity being used.

25. A processing system usable as a storage server node in a network storage cluster that includes a plurality of storage server nodes, the processing system comprising:

a processor;

a network interface, coupled to the processor, through which to communicate with a remote processing system over a network; and

a memory accessible to the processor and storing code which embodies;

a gatherer to gather statistics relating to CPU utilization and throughput for each of a set ports in the network storage cluster and to distribute the statistics to another storage server node in the cluster;

a virtual interface manager to determine a set of weights for a set of network addresses of the cluster, based on the distributed statistics and to provide the weights as input for forming a weighted list of available network addresses; and

a domain name server (DNS) to respond to a client name service request, by randomly indexing into the weighted list to resolve the client name service request to a network address in the storage cluster.

26. A processing system as recited in claim 25, wherein each of the storage server nodes determines the same set of weights for the set of network addresses.

27. A processing system as recited in claim 25, wherein the DNS resolves the client name service request to load balance between different ports on a particular storage server node in the network storage cluster.

28. A processing system as recited in claim 27, wherein the set of weights are determined by, for each of a plurality of network addresses of the storage cluster:

computing a first value corresponding to a percentage of CPU capacity being used;

computing a second value corresponding to a percentage of throughput capacity being used; and

computing a third value as a function of the first value and the second value.