Distributing lookup operations in a striped storage system

• US 7,996,607 B1
• Filed: 01/28/2008
• Issued: 08/09/2011
• Est. Priority Date: 01/28/2008
• Status: Active Grant

• Current Assignee: NetApp, Inc.
• Original Assignee: NetApp, Inc.
• Inventors: Jernigan, Richard P. IV, Thoppai, Omprakaash
• Primary Examiner(s): Gu; Shawn X
• Application Number:
  US12/020,949
• Time in Patent Office: 1,289 Days
• Field of Search: None
• US Class Current: 711/113
• CPC Class Codes: G06F 12/0866 for peripheral storage systems, e.g. disk cache
• CPC Class Codes: G06F 2212/262 configured as RAID
• CPC Class Codes: G06F 2212/466 Metadata, control data
• CPC Class Codes: H04L 67/561 Adding application-functional data or data for application control, e.g. adding metadata
• CPC Class Codes: H04L 67/5682 Policies or rules for updating, deleting or replacing the stored data

Abstract

The present invention provides a mechanism for storing and accessing attributes of a data container (e.g., characteristics such as a name of a data container used to perform a lookup operation against the data container) in cache memory within storage systems in a cluster of storage systems. Further, a mechanism is provided so that an attribute of a data container is inserted into a cache memory of preferably one storage system (in the cluster) that caches metadata of the data container (e.g., information describing the data container). As a result, a lookup operation for a data container is directed to and served by the storage system that caches the metadata and the attribute of the data container.

Claims

1. A method for distributing lookup operations in a cluster comprising one or more N-blades connected to one or more D-blades, the method comprising:

maintaining a D-blade storing, at one or more storage devices, mappings between a data container attribute that defines a data container and an inode number associated with the data container;

at the D-blade storing the mappings, performing the following steps;

determining an identification (ID) of a D-blade in the cluster for caching the data container attribute on a storage device to which a lookup request for the data container will be directed by an N-blade, and

selecting an inode number for the data container attribute so that a D-blade that serves a lookup request has a same ID as a D-blade that serves a metadata request;

at a D-blade selected to cache metadata and the data container attribute, is performing;

in response to receiving a metadata request from the N-blade, caching the metadata and the data container attribute and providing the metadata to a client,

in response to receiving, from the N-blade, the lookup request for the data container based on the data container attribute, satisfying the received lookup request without the lookup request for the data container being sent by the N-blade to the D-blade that stores the mappings.

2. The method of claim 1, wherein the N-blade is configured to receive requests from clients and to communicate with one or more D-blade.

3. The method of claim 1, wherein each of the one or more D-blades are configured to communicate with one or more N-blade and are connected to storage devices.

4. The method of claim 1, wherein the data container attribute that defines the data container is a name of the data container.

5. The method of claim 1, wherein the data container attribute is infrequently changing data and wherein the metadata represents data that is changed more frequently than infrequently changing data.

6. The method of claim 1, wherein the ID of the D-blade caching the data container attribute is determined by applying a first function to the data container attribute, and wherein the inode number for the data container attribute is selected so that upon application of a second function on the inode number for the data container attribute, a value is selected equal to the determined ID so that the received lookup request and the request for metadata are directed to the same D-blade without being sent to the D-blade storing the mappings.

7. The method of claim 1, wherein after receiving the lookup request for a data container based on the date container attribute, performing the following steps by a node caching the requested data container attribute:

determining whether metadata for the data container is cached at the node; and

in response to the metadata for the data container not being cached, avoiding deleting the requested data container attribute from a cache.

8. The method of claim 1, wherein an inode number is an index of an inode data structure configured to store metadata about the data container.

9. The method of claim 1, wherein the data container comprises a file.

10. A method for distributing lookup operations in a cluster of nodes, the method comprising:

maintaining a D-blade storing, on one or more storage devices, mappings between a data container attribute and an inode number associated with a data container;

distributing a cache of the data container attribute that defines a data container in the cluster so that the data container attribute is cached by one D-blade in the cluster comprising one or more N-blades connected to one or more D-blades;

receiving, at the D-blade storing the mappings, a request to create a data container, from an N-blade, and creating a data container based on the data container attribute;

determining, at the D-blade storing the mappings, using a first function, an identification (ID) of the D-blade in the cluster to cache the data container attribute on a storage device and to which a lookup request for a data container will be directed by an N-blade that will receive a lookup request from a client;

allocating, at the D-blade storing the mappings, an inode number for the data container attribute so a lookup request and a request for metadata are directed to the D-blade that caches the data container attribute;

caching, by the D-blade having the ID, the data container attribute and metadata;

determining using the first function, by an N-blade receiving a lookup request for the data container based on the a data container attribute, the ID of the D-blade that caches the data container attribute and forwarding the received lookup request to the D-blade that caches the data container attribute without sending the received lookup request to the D-blade that stores the mappings.

11. A non-transitory computer-program product comprising:

a computer-readable medium having computer program code embodied thereon for distributing lookup operations in a cluster comprising one or more N-blades connected to one or more D-blades, the computer program code configured to;

maintain a D-blade storing, at one or more storage devices, mappings between a data container attribute that defines a data container and an inode number associated with the data container;

at the D-blade storing the mappings, configured to;

determine an identification (ID) of a D-blade in the cluster for caching the data container attribute on a storage device to which a lookup request for the data container will be directed by an N-blade, and

select an inode number for the data container attribute so that a D-blade that serves the lookup request has a same ID as a D-blade that serves a metadata request; and

is at a D-blade selected to cache metadata and the data container attribute, configured to;

in response to receiving a metadata request from an N-blade of the one or more N-blades, cache the metadata and the data container attribute and providing the metadata to a client, and

in response to receiving, from an N-blade of the one or more N-blades, a lookup request for the data container based on the data container attribute, satisfying the received lookup request without sending the received lookup request from an N-blade to the D-blade that stores the mappings