Patent Application
20160140003
Embodiments of the disclosed subject matter generally relate to the field of storage networks and, more particularly, to storage controller replacement within cross-cluster redundancy configurations.
Computer clusters implement a form of distributed computing. A computer cluster consists of a set of nodes that are configured and communicatively coupled in a cooperative manner to share resources and in some respects operate as a single system. The components of a cluster typically include multiple server nodes and one or more cluster management nodes interconnected by a local area network (LAN), with each node running its own instance of a common operating system. Clusters are usually deployed to improve performance and availability over that of centralized computing, while typically being more cost-effective than single computers of comparable speed or availability.
A storage cluster is a type of networked computer cluster generally characterized as including multiple interconnected storage nodes. Each storage node consists of a controller coupled to a mass storage unit such as an array of storage disks or solid state drives (SSDs) on which data, sometimes referred to as “backend data,” is stored. The storage node controller performs server-like functions for optimizing access to and usage of storage resources including the stored data. The mass storage unit may be a Redundant Array of Independent Disks (RAID) that provides long-term, non-volatile data storage.
Ensuring continuous, uninterrupted access to backend data is a vital function of most storage clusters. So-called High Availability (HA) storage is often used to ensure uninterrupted access to backend data in the event of an interruption to a given storage node's operation. The interruption may be due to a hardware or software failure, or due to maintenance (e.g., replacement) of a storage node. An HA configuration may define a cluster (an HA cluster) or may be a cluster configuration feature such as one or more HA pairs within an otherwise defined cluster. In either case, the basic HA storage configuration consists of at least two somewhat independent storage nodes that perform mutual backup roles under the management of system control code and related configuration settings. Simply, when one of the nodes fails, the other immediately assumes control of its HA partner node's operation and storage.
The increasing scale of distributed data storage has raised the need to expand protection of stored data and uninterrupted access thereto beyond intra-cluster backup redundancy. This need is being addressed by the growing prevalence of data redundancy across clusters. Storage redundancy across clusters, such as within data centers which may be physically separated by tens or even hundreds of kilometers, uses data replication such as by data mirroring. In this manner, the data and uninterrupted access thereto are protected against site-wide failures that may result, for example, from loss of power.
The embodiments may be better understood by referencing the accompanying drawings.
The description that follows includes example systems, methods, techniques, instruction sequences and computer program products that embody techniques of the disclosed subject matter. However, it is understood that the described embodiments may be practiced without one or more of these specific details. Well-known instruction instances, protocols, structures and techniques have not been shown in detail in order not to obfuscate the description. As utilized herein, the term “node” with or without additional descriptors (e.g., storage node, controller node, etc.) may refer to a cluster configuration identifier or to a controller device such as that depicted in
Clustered storage configured as HA pairs is widely used to provide distributed storage while protecting against system hardware and/or software failures in the member devices. Node takeover (often referred to as “failover”) is a standard mechanism used by HA pairs to provide uninterrupted access to stored data upon a failure or abnormal termination of a storage node. Essentially, a failover entails a backup node assuming control of the storage devices and network connections previously controlled by the failed node so that the backup node can provide uninterrupted data availability.
Cross-cluster data redundancy is designed to protect against larger scale failures, such as rolling hardware failures, and is sometimes referred to in terms of “disaster recovery” redundancy or “site recovery” redundancy. Synchronous data mirroring and non-volatile random access memory (NVRAM) write cache replication may be used to maintain a consistent operational state and version of aggregate data between two sites that, unlike nodes local to a given cluster, do not share data storage devices. In one embodiment, the disclosure is directed to maintaining HA operational continuity for a controller replacement within a cross-cluster redundancy configuration.
Information storage on each of storage arrays 125a and 125b is preferably implemented as one or more addressable storage areas, referred to as storage volumes, that reside on a collection of physical storage drives 115a-m and 115b-n cooperating to define an overall logical arrangement of volume block number space on the volume(s). Each logical volume is generally, although not necessarily, associated with its own file system. The storage drives within a logical volume/file system may comprise any combination of solid state drives (SSDs) and/or magnetic disks and are typically organized as one or more groups, wherein each group may be operated as a Redundant Array of Independent Disks (RAID).
As further depicted in
The clients 102 may be general-purpose computers configured to interact with nodes 105a and 105b in accordance with a client/server model of information delivery. That is, clients 102 may request the services of a node either directly or via a host server, and the node responds to the client service requests by exchanging packets over network 116. A client may issue packets including file-based access protocols, such as the Common Internet File System (CIFS) protocol or Network File System (NFS) protocol, over the Transmission Control Protocol/Internet Protocol (TCP/IP) when accessing information in the form of files and directories. Alternatively, the client may issue packets including block-based access protocols, such as the Small Computer Systems Interface (SCSI) protocol encapsulated over TCP (iSCSI) and SCSI encapsulated over Fibre Channel (FCP), when accessing information in the form of blocks.
The purpose of an HA cluster pair such as HA pair 100 is to provide operational continuity and uninterrupted storage resources availability during maintenance events such as controller replacement as well as unexpected events such as hardware or software failures. During normal cluster operation, a storage controller node, such as node 105a that is connected to and has primary operational control of storage drives 115a-m is identified in accordance with the HA configuration as the “home” of drives 115a-m. As such HA node 105a is also identified as the “current owner” at initialization and is primarily responsible for servicing data requests directed to blocks of volumes contained on storage drives 115a-m. Similarly, storage node 105b is primarily responsible for the SSDs and/or disks represented as storage drives 115b-n within storage array 125b. HA pair 100 is configured such that either of nodes 105a or 105b can take over data servicing capabilities for the other node in the event of a failure or maintenance downtime. As used herein, “takeover” may refer to either a planned takeover of one HA node partner by the other partner, or a “failover” sequence in which the takeover is caused by a runtime failure or other unexpected event.
An HA partner node is able to take over the management services of the other controller node by virtue of shared access to storage devices and write cache replication between the nodes. In the depicted embodiment, nodes 105a and 105b have shared operational access to storage arrays 125a and 125b. Furthermore, HA pair 100 is configured such that NVRAM write cache content (depicted and described with reference to
As depicted and explained in further detail with reference to
Node 105 is further depicted as a dual processor controller executing a storage operating system 206 that preferably implements a high-level module, such as a file system, to logically organize the information as a hierarchical structure of named data containers, such as directories, files and special types of files called virtual disks (sometimes referred to as “blocks”) on the SSDs or disks. However, it will be apparent to those of ordinary skill in the art that node 105 may alternatively comprise a single or more than two processor system. In one embodiment, one processor 202a executes the functions of the N-module 104a on the node, while the other processor 202b executes the functions of the D-module 106a.
Memory 210 comprises storage locations that are addressable by the processors and adapters for storing software program code and data structures associated with the disclosed embodiments. The processor and adapters may, in turn, comprise processing elements and/or logic circuitry configured to execute the program code and manipulate the data structures. Storage operating system 206, portions of which are typically resident in memory and executed by the processing elements, functionally organizes node 105 by, inter alia, invoking storage operations in support of the storage service implemented by the node. It will be apparent to those skilled in the art that other processing and memory means, including various computer readable media, may be used for storing and executing program instructions pertaining to the embodiments described herein. In the depicted embodiment, storage operating system 206 further includes a cross-cluster redundancy management module 207 that includes program instructions that when executed by one or more processors implement one or more of the functions depicted and described with reference to
Network adapter 208 comprises a plurality of ports adapted to couple node 105 to one or more clients 102 over point-to-point links, wide area networks, virtual private networks implemented over a public network or a shared local area network. Network adapter 208 thus may comprise the structure and circuitry as well as the logic constructs needed to communicatively couple node 105 to the network 116 (
Storage adapter 218 functions cooperatively with storage operating system 206 to access storage resources (e.g., requested information) within storage arrays 125a and 125b on behalf of clients 102. Information may be stored in storage arrays 125a and 125b on any type of attached array of writable storage device media such as magnetic tape, optical media, electronic random access memory, SSD, and any other similar media adapted to store information, including data and meta data. In the depicted embodiment, the information is stored on storage drives 115 of storage arrays 125a and 125b. Storage adapter 218 further includes multiple ports having I/O interface circuitry (not depicted) communicatively coupled to drives 115 over an I/O interconnect, such as a FC link topology.
As explained with reference to
Cross-cluster configuration 350 includes HA pair A1/A2 that share operational access to storage arrays 306a and 306b through a Fibre Channel (FC) switching network comprising FC switches 302 and 304. Configuration 350 includes a corresponding “partner” HA pair B1/B2 that share operational access to storage arrays 316a and 316b through an FC switching network comprising FC switches 312 and 314. Cross-cluster configuration 355 includes HA pair A3/A4 (comprising storage controller nodes 325a and 325b) that share operational access to storage arrays 326a and 326b through an FC switching network comprising FC switches 322 and 324. Cross-cluster configuration 355 also includes HA pair B3/B4 (comprising storage controller nodes 335a and 335b) that share operational access to storage arrays 336a and 336b through an FC switching network comprising FC switches 332 and 334.
The configuration and operation of configurations 350 and 355 are substantially similar. The intra-cluster HA operations and cross-cluster redundancy operations of configuration 350 will be described for purposes of illustration. During normal runtime operations, each of controller nodes A1305a and A2305b controls its own respectively assigned subset of storage arrays 306a and 306b. For example, HA pair A1/A2 may be configured within Cluster A such that controller node 305a has operational control of storage array 306a via switch 302. The shared access to storage arrays 306a and 306b, together with the replication of NVRAM I/O staging cache content (not depicted in
HA pairs A1/A2 and B1/B2 are mutually programmed and otherwise configured within configuration 350 to provide cross-cluster data redundancy and operational continuity. Each node has a configuration-specified relationship to each of its two cross-cluster partners in addition to its intra-cluster HA partnership. In the depicted example, node A1 may be associated by the configuration to have a cross-cluster “partner” relation with node B1 and is further associated by the configuration to have an “auxiliary” (i.e., backup) cross-cluster partner relationship with node B2. In such a case, node B1 would be the reciprocal cross-cluster partner of node A1, and node A2 would be the auxiliary cross-cluster partner of node B1. In the depicted embodiment, HA node pairs A3/A4 and B3/B4 may be similarly configured to form cross-cluster partners and auxiliary partners within configuration 355.
Similar to the HA pair mechanism described above, the operational continuity provided by the respective cross-cluster redundancy partnerships and auxiliary partnerships is enabled in part by synchronously replicating NVRAM I/O staging cache content between the respective partners. Data may be replicated between the cross-cluster partners, such as nodes A1 and B1, by synchronously mirroring stored data aggregates between the respective backend storage devices 306a, b and 316a, b via long haul FC connections 342 and 344. In one embodiment, RAID level data mirroring may be used to perform the cross-cluster backend storage replication. As shown in the depicted embodiment, controller nodes 305a, 305b, 315a, and 315b respectively include cross-cluster redundancy management modules 362a, 362b, 364a, and 364b. As depicted and explained in further detail with reference to
The A2 node controller 305b may include one or more components depicted in
Node A2305b announces its reentry onto the FC interconnect of cross-cluster configuration 350 by multi-casting a node advertisement that indicates node A2's identity. As shown at step 412, the node advertisement multi-cast preferably coincides with a giveback wait phase during which A2 does not yet have access to its resources such as a mailbox root disk within storage array 306b that records A2's controller device ID and cross-cluster configuration ID. At this point, before node A2 has access to its unique cross-cluster configuration ID, the process continues at step 414 with A2 multi-casting its node advertisement to all member nodes (i.e., A1, B1, and B2) of cross-cluster configuration 350. In accordance with the depicted embodiment, the node advertisement includes node identification information including the switch identifiers that A2 obtained at step 410.
Step 408 depicts node A1 receiving any given node advertisement from any given member node of cross-cluster configuration 350. Typically, the node advertisement will include a Cluster ID, a Node System ID, and a Node Config ID. The Cluster ID identifies the cluster that the node belongs to such that in the depicted embodiment, nodes A1 and A2 will have the same Cluster ID as will B1 and B2. The Node System ID is a number or alphanumeric code that individually identities the controller hardware. In the depicted example, the Node System ID will have changed for node A2 because its controller has been replaced. The Node Config ID is a code that identifies the node in terms of its function, connectivity, and mutual relations with other nodes in the cross-cluster configuration. Therefore, the Node Config ID remains the same in association with a given “node” regardless of whether or not the node's controller is replaced.
Given the four-node membership of cross-cluster configuration 350, any of the member nodes such as A1 as shown at step 408 may receive anode advertisement message from any of at least three member nodes. As explained above, all member nodes have a defined role with respect to each of the other members and the Node Config ID defines this role. However, when a node such as A2 is restarted following controller replacement, it does not have access to its Node Config ID which is stored on a mailbox root disk, access to which has yet to be returned by node A1. During the giveback wait phase of startup for A2, node A1 receives a multi-cast from node A2 (step 408) that includes the switch identifier information obtained by node A2 at step 410. As shown at step 416, node A1 determines whether the intra-cluster identifier switch identifier in this case) included with the multi-cast from node A2 matches or otherwise corresponds to its own intra-cluster connectivity information to determine whether A2 is its HA pair partner. In response to determining that the switch identifiers contained in A2's node advertisement message match its own intra-cluster connectivity information (e.g., matches its own corresponding record of which local switches it is connected to), A1 commences giveback of the storage resources (step 418) and the process ends with A2 being brought back online within the cross-cluster configuration.
The Node System ID corresponds to the device identifier of a node's controller, typically represented as the serial number of the NVRAM card contained within the controller. In the depicted example, the copy of the cross-cluster configuration information stored locally within A2 will include an identifier of the controller that was replaced as the Node System ID. As shown at step 508, A2 compares the locally recorded Node System ID with an identifier of its newly installed controller device (step 510). If, as shown beginning at step 512, A2 determines that the locally stored controller identifier is different than the identifier of the newly installed controller, A2 generates a controller replacement message to be multi-cast (step 514) to the other cross-cluster configuration members.
The controller replacement message includes the identifier of the replaced A2 controller (i.e., the current locally stored Node System ID for A2) and the corresponding identifier of the replacement A2 controller, both in association with A2's Node Config ID, which, as previously explained, remains unchanged following controller replacement. Having received the multi-cast controller replacement message, each of recipient member nodes A1, B1, and B2 uses the specified Node Config ID as a key to identify which member node's Node System ID requires replacement (step 516). Upon replacing the Node System ID (i.e., controller device ID) in its locally stored copy of the cross-cluster configuration information as shown at step 518, each member node sends an ACK to the multi-cast sender, A2. The process is completed when A2, having confirmed receipt of ACK replies from all member nodes updates its own local configuration copy to replace the previous controller ID with the new controller ID (steps 520, 522, and 524).
As will be appreciated by one skilled in the art, aspects of the disclosed subject matter may be embodied as a system, method or computer program product. Accordingly, embodiments of the disclosed subject matter may take the form of an entirely hardware embodiment, an entirely software embodiment (including firmware, resident software, micro-code, etc.) or an embodiment combining software and hardware aspects that may all generally be referred to herein as a “circuit,” “module” or “system.” Furthermore, embodiments of the disclosed subject matter may take the form of a computer program product embodied in one or more computer readable medium(s) having computer readable program code embodied thereon.
Any combination of one or more computer readable medium(s) may be utilized. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Computer program code for carrying out operations for aspects of the disclosed subject matter may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like and conventional procedural programming languages, such as the “C” programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider).
While the embodiments are described with reference to various implementations and exploitations, it will be understood that these embodiments are illustrative and that the scope of the disclosed subject matter is not limited to them. In general, techniques for non-disruptively replacing a storage controller as described herein may be implemented with facilities consistent with any hardware system or hardware systems. Many variations, modifications, additions, and improvements are possible.
