Data storage system for a multi-client network and method of managing such system

• US 20050154841A1
• Filed: 03/07/2005
• Published: 07/14/2005
• Est. Priority Date: 05/08/2001
• Status: Abandoned Application

• Current Assignee: Maximum Throughput, Incorporated (Avid Technology Incorporated)
• Original Assignee: Maximum Throughput, Incorporated (Avid Technology Incorporated)
• Inventors: Xu, Yue, Zhou, Xinliang, McCauley, Steeve, Rajekar, Ashutosh, Findleton, Lain B., Rosenblatt, Ariel, Sastri, Gautham
• Application Number:
  US11/073,953
• Publication Number:
  US 20050154841A1
• Time in Patent Office: Days
• US Class Current: 711/148
• CPC Class Codes: G06F 2003/0697 device management, e.g. handlers, drivers, I/O schedulers
• CPC Class Codes: G06F 3/0601 Interfaces specially adapted for storage systems

Abstract

The data storage system comprises a scalable number of routing processors (RPs) through which clients of a network communicate. The storage system also includes a scalable number of storage processors (SPs) connected to a scalable number of storage units (SUs). This data storage system provides a new and hybrid approach which lies in between conventional NAS and SAN environments. It creates a unified and scalable storage pool accessible through a single consistent directory without the need for a metadata controller (MDC). There is thus no table lookup at a central node and no single point of failure. It allows a dissociation of the relationship between the physical path and the actual location where the data objects are stored.

Claims

1. A method of processing operation requests related to data objects in a data storage system connected to a multi-client network, the data storage system comprising a storage pool having a plurality of storage units (SUs), the method comprising:

providing at least one routing processor (RP) and a plurality of storage processor (SPs) coupled to the RP and the SUs;

dividing the storage pool into logical containers and assigning each logical container to one of the SPs;

at the RP, receiving an operation request related to a data object from a client of the network;

determining which one of the containers corresponds to the data object;

sending the operation request to the SP assigned to the corresponding logical container;

receiving the operation request at the assigned SP; and

processing the operation request at the SP.

2. A method according to claim 1, wherein the method comprises:

sending the data object with the corresponding requested operation.

3. A method according to claim 1, further comprising:

providing a management station (MS) interconnected to the RP and each SP;

monitoring the operation of at least each SP; and

in case of a failure of one of the SPs, reassigning logical containers of the failed SP to at least one of the other SPs.

4. A method according to claim 3, wherein the act of reassigning logical containers comprises:

updating a configuration database provided in the RP and each SP to reflect new logical container assignations.

5. A method according to claim 1, further comprising:

sending data objects between the SPs and the SUs through a high-speed switch.

6. A method according to claim 5, wherein the high-speed switch is a Fiberchannel switch.

7. A method according to claim 1, further comprising:

verifying at the RP if the operation request is successfully completed within a maximum delay; and

sending a corresponding notification to the client.

8. A method of processing operation requests associated with data objects in a data storage system connected to a multi-client network, the data storage system comprising a storage pool having a plurality of storage units (SUs) divided into logical containers, each logical containers being assigned to one among a plurality of storage processors (SPs), the method comprising:

receiving at a routing processor (RP) a save request from a client of the network concerning a new data object;

determining, from at least one attribute of the new data object, a destination container among the logical containers for storing the new data object;

sending the new data object to the SP to which the selected container is assigned;

receiving the new data object at the SP handling the destination container; and

storing the new data object in the storage pool at the destination container.

9. A method according to claim 8, further comprising:

sending data indicative of a result of the save request to the client from which it originates.

10. A method according to claim 8, wherein the destination container is selected using a scheme carrying out a statistically substantially-uniform distribution of new data objects among containers, the scheme outputting a number corresponding to the destination container in which the new data object is to be stored.

11. A method according to claim 10, wherein the scheme comprises a convolution algorithm.

12. A method according to claim 11, wherein the convolution algorithm comprises the act of generating a number using a Cyclic redundancy check (CRC) algorithm and applying a mask thereto.

13. A method according to claim 8, further comprising:

sending the new data object between the SP and one of the SUs of the storage pool through a high-speed switch.

14. A method according to claim 13, wherein the high-speed switch is a Fiberchannel switch.

15. A method of routing new data objects in a data storage system connected to a multi-client network, the data storage system having a storage pool divided in a predetermined number of logical containers in which data objects are stored, each data object including contents and at least one attribute, the method comprising:

selecting one of the logical containers as a destination container to store a new data object received from a client of the network, the destination container being selected using a scheme providing a statistically substantially uniform distribution of the data objects between the logical containers using at least one attribute of each data object; and

sending the new data object to the destination container.

16. A method according to claim 15, further comprising:

verifying at the RP if the new data object is successfully stored in the destination container within a maximum delay; and

sending a corresponding notification to the client.

17. A data storage system for storing data objects, the data storage system being connected to a multi-client network and being provided with a storage pool having a plurality of storage units (SUs), the system comprising:

at least one routing processor (RP) coupled to the network;

a plurality of storage processors (SPs) coupled to the RP;

a storage pool having a plurality of storage units (SUs), the storage pool being divided into logical containers;

a switch to interconnectivity couple the SPs and the SUs; and

a managing station (MS) coupled to the RP and the SPs, the MS maintaining a main configuration database and corresponding configuration databases in the RP and the SPs to indicate which of the SPs is being assigned to each logical container.

18. A data storage system according to claim 17, wherein the MS is coupled to the RP and the SPs by an independent control network.

19. A data storage system according to claim 17, wherein the switch is a Fiberchannel switch.

20. A data storage system according to claim 17, wherein more than one RP is provided, each of the RPs being coupled to the SPs by a router.

21. A data storage system according to claim 17, wherein each RP comprises:

means for verifying if an operation request concerning a data object is successfully completed within a maximum delay; and

means for sending a corresponding notification to a client of the network from which the operation request originated.

22. A data storage system according to claim 17, wherein each RP comprises:

means for selecting one of the logical containers as a destination container to store a new data object, the means using a scheme providing a statistically substantially-uniform distribution of the data objects between the containers from at least one attribute of each data object.

23. A data storage system according to claim 22, wherein means for selecting one of the logical containers as a destination container comprises:

means for generating a number using a Cyclic redundancy check (CRC) algorithm; and

means for applying a mask to obtain a number indicative of the destination container.