Network topology for a scalable data storage system

• US 20070214105A1
• Filed: 03/08/2006
• Published: 09/13/2007
• Est. Priority Date: 03/08/2006
• Status: Abandoned Application

• Current Assignee: Harmonic Incorporated
• Original Assignee: Omneon Video Networks
• Inventors: Sfarti, Adrian, Wanigasekara-Mohotti, Don, Craig, Donald
• Application Number:
  US11/371,678
• Publication Number:
  US 20070214105A1
• Time in Patent Office: Days
• US Class Current: 1/1
• 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

A data storage system has a number of server groups, where each group has data storage servers. A file is stored in the system by being spread across two or more of the servers. The servers are communicatively coupled to internal packet switches. An external packet switch is communicatively coupled to the internal packet switches. Client access to each of the servers is through one of the internal packet switches and the external packet switch. Other embodiments are also described and claimed.

Claims

1. A data storage system comprising:

a first plurality of server groups, each group having a plurality of data storage servers, wherein a file is stored in the system by being spread across two or more of the data storage servers of said groups;

a first plurality of internal packet switches to which the data storage servers of the first plurality of server groups are communicatively coupled, respectively; and

a first external packet switch that is communicatively coupled to the plurality of internal packet switches, wherein client access to each of the data storage servers is through one of the internal packet switches and the external packet switch.

2. The system of claim 1 wherein the data storage servers in each group comprise a metadata server and a plurality of content servers, and the file is to be stored in the system by being spread across two or more of the content servers in the system as determined by a metadata server.

3. The system of claim 2 wherein the data storage servers in each group communicate, at a physical layer, with a respective one of the internal packet switches and not the first external packet switch.

4. The system of claim 3 further comprising a second external packet switch that is communicatively coupled to the plurality of internal packet switches, to provide redundant client access to each of the data storage servers through one of the internal packet switches.

5. The system of claim 4 further comprising a plurality of adapter switches each being communicatively coupled between a) the data storage servers of a respective one of the server groups and b) a respective one of the internal packet switches, each adapter switch having a) a plurality of low bandwidth ports coupled to the data storage servers of the respective server group and b) a plurality of high bandwidth ports coupled to the respective internal packet switch.

6. The system of claim 1 further comprising a plurality of adapter switches each being communicatively coupled between a) the data storage servers of a respective one of the server groups and b) a respective one of the internal packet switches, each adapter switch having a) a plurality of low bandwidth ports coupled to the data storage servers of the respective server group and b) a plurality of high bandwidth ports coupled to the respective internal packet switch.

7. The system of claim 1 wherein each of the server groups comprises a separate enclosure containing its own power supply and fan.

8. The system of claim 7 wherein the separate enclosure is a server rack.

9. A data storage system comprising:

a first plurality of clusters, each cluster having a plurality of server groups and an internal packet switch, each server group in a duster having a plurality of content servers and a metadata server communicatively coupled to the internal packet switch of the cluster, wherein a file is stored in the system by being spread across two or more content servers as determined by a metadata server; and

a first plurality of external packet switches that are communicatively coupled to the plurality of clusters, respectively, via the internal packet switch of each cluster.

10. The system of claim 9 wherein the external packet switches are Ethernet switches.

11. The system of claim 9 wherein the external packet switches are Infiniband switches.

12. The system of claim 9 wherein each of the server groups is in a separate enclosure containing its own power supply and fan.

13. The system of claim 12 wherein the separate enclosure is a server rack.

14. The system of claim 13 wherein each of the external packet switches is in a separate enclosure containing its own power supply, processor, memory, ports and packet forwarding table.

15. The system of claim 9 further comprising a second plurality of external packet switches that are communicatively coupled to the plurality of clusters, respectively, via the internal packet switch of each cluster.

16. The system of claim 9 wherein each of the external packet switches has N ports half of which are coupled to the plurality of clusters, respectively, and half of which are available.

17. The system of claim 16 further comprising a second plurality clusters, each cluster has N/2 ports that are communicatively coupled to the available ports of the external packet switches, respectively.

18. A method for providing a scalable data storage system, comprising:

providing a data storage system having a plurality of existing server groups each group having a plurality of data storage servers, wherein a file is stored in the system by being spread across two or more of the data storage servers of said groups, a plurality of existing internal packet switches to which the data storage servers of the plurality of existing server groups are communicatively coupled, respectively, and an existing external packet switch that is communicatively coupled to the plurality of internal packet switches, wherein client access to each of the data storage servers is through one of the internal packet switches and the external packet switch;

providing a plurality of upgrade clusters each upgrade cluster having an upgrade internal packet switch; and

connecting a plurality of ports of the upgrade internal packet switches belonging to the upgrade clusters to a plurality of available ports of the existing external packet switch, respectively.

19. A method for providing a scalable data storage system, comprising:

providing a data storage system having a cluster, the cluster having a plurality of server groups and an internal packet switch, each server group having a plurality of content servers and a metadata server communicatively coupled to the internal packet switch, wherein a file is stored in the system by being spread across two or more content servers as determined by the metadata server, and an external packet switch that is communicatively coupled to the cluster via the internal packet switch;

providing an upgrade server group;

replacing the existing external packet switch with one that has a greater number of ports;

replacing the existing internal packet switch with one that has a greater number of ports; and

merging the upgrade server group with the cluster.

20. The method of claim 19 wherein the number of ports in the replacement internal packet switch that connect with the replacement external switch is twice the number of ports in the existing internal packet switch that were connected with the existing external switch.

21. A method for providing a scalable data storage system, comprising:

providing a plurality of existing server groups, each group having a plurality of data storage servers, wherein a file is stored in the system by being spread across two or more of the data storage servers of said groups;

providing a pair of existing external switches each with 2N ports;

providing a plurality of internal packet switches to which the data storage servers of said server groups are communicatively coupled, respectively, the internal packet switches collectively having 2N ports which are connected to the 2N ports of the existing external switches by existing links;

providing a plurality of upgrade server groups communicatively coupled to a plurality of upgrade internal switches, and at least two upgrade external switches, each with 2N ports; and

disconnecting the existing links to N ports of each of the existing external switches, and reconnecting them to ports of the upgrade external switches.

22. The method of claim 21 further comprising:

connecting a plurality of ports in each of the upgrade internal switches to the existing and upgrade external switches.