Symantec Operating Corporation

Netapp, Inc.

Symantec Corporation

John A Muth

Gregory L Slaughter

Oleg Kiselev

Thomas E Saulpaugh

Stephen H Strange

Dhaval Patel

Craig Harmer

Edward D McClanahan

Andrew K Krumel

Manish Swaminathan

Kayuri Hasmukh Patel

Steven S Watanabe

Thomas Lanzatella

Bob Schatz

Kimberly A Malone

Paul Czarnik

Angshuman Bezbaruah

Milind Vithal Borate

Paul L Borrill

Chaitanya V Patel

Kai Chiu Wong

Juan C Gomez

Srinath Krishnamachari

Marc P Kwiatkowski

Girish Jorapurkar

A third vote consensus technique enables a first node, i.e., a surviving node, of a two-node cluster to establish a quorum and continue to operate in response to failure of a second node of the cluster. Each node maintains configuration information organized as a cluster database (CDB) which may be changed according to a consensus-based protocol. Changes to the CDB are logged on a third copy file system (TCFS) stored on a local copy of TCFS (L-TCFS). A shared copy of the TCFS (i.e., S-TCFS) may be stored on shared storage devices of one or more storage arrays coupled to the nodes. The local copy of the TCFS (i.e., L-TCFS) represents a quorum vote for each node of the cluster, while the S-TCFS represents an additional 'tie-breaker' vote of a consensus-based protocol. The additional vote may be obtained from the shared storage devices by the surviving node as a third vote to establish the quorum and enable the surviving node to cast two of three votes (i.e., a majority of votes) needed to continue operation of the cluster. That is, the majority of votes allows the surviving node to update the CDB with the configuration information changes so as to continue proper operation of the cluster.

Third vote consensus in a cluster using shared storage devices

A high availability (HA) failover manager maintains data availability of one or more input/output (I/O) resources in a cluster by ensuring that each I/O resource is available (e.g., mounted) on a hosting node of the cluster and that each I/O resource may be available on one or more partner nodes of the cluster if a node (i.e., a local node) were to fail. The HA failover manager (HA manager) processes inputs from various sources of the cluster to determine whether failover is enabled for a local node and each partner node in an HA group, and for triggering failover of the I/O resources to the partner node as necessary. For each I/O resource, the HA manager may track state information including (i) a state of the I/O resource (e.g., mounted or un-mounted); (ii) the partner node(s) ability to service the I/O resource; and (iii) whether a non-volatile log recording I/O requests is synchronized to the partner node(s). The HA manager interacts with various layers of a storage I/O stack to mount and un-mount the I/O resources on one or more nodes of the cluster through the use of well-defined interfaces, e.g., application programming interfaces.

High availability failover manager

A layout of a file system is optimized to meet storage requirements of a distributed consensus protocol implemented on a plurality of nodes of a cluster. Illustratively, the file system has an on-disk layout representation that enables efficient input/output (I/O) operation performance. The on-disk layout of the file system embodies a plurality of file system objects: membership, snapshot and log objects. Each object is allocated an area or region (e.g., a contiguous storage space) on-disk. In addition, each object has two or more storage container instances, e.g., files. In the case of snapshot and membership objects, the two files of each object are used in a circular (alternating) fashion so that write operations directed to a first file storing a current copy of the snapshot/membership content may be performed without destroying a previous committed copy of the snapshot/membership content stored in a second file.

Optimized file system layout for distributed consensus protocol

A rate matching technique may be configured to adjust a rate of cleaning of one or more selected segments of the storage array to accommodate a variable rate of incoming workload processed by a storage input/output (I/O) stack executing on one or more nodes of a cluster. An extent store layer of the storage I/O stack may clean a segment in accordance with segment cleaning which, illustratively, may be embodied as a segment cleaning process. The rate matching technique may be implemented as a feedback control mechanism configured to adjust the segment cleaning process based on the incoming workload. Components of the feedback control mechanism may include one or more weight schedulers and various accounting data structures, e.g., counters, configured to track the progress of segment cleaning and free space usage. The counters may also be used to balance the rates of segment cleaning and incoming I/O workload, which may change depending upon an incoming I/O rate. When the incoming I/O rate changes, the rate of segment cleaning may be adjusted accordingly to ensure that rates are substantially balanced.

Rate matching technique for balancing segment cleaning and I/O workload

An optimized segment cleaning technique is configured to efficiently clean one or more selected portions or segments of a storage array coupled to one or more nodes of a cluster. A bottom-up approach of the segment cleaning technique is configured to read all blocks of a segment to be cleaned (i.e., an 'old' segment) to locate extents stored on the SSDs of the old segment and examine extent metadata to determine whether the extents are valid and, if so, relocate the valid extents to a segment being written (i.e., a 'new' segment). A top-down approach of the segment cleaning technique obviates reading of the blocks of the old segment to locate the extents and, instead, examines the extent metadata to determine the valid extents of the old segment. A hybrid approach may extend the top-down approach to include only full stripe read operations needed for relocation and reconstruction of blocks as well as retrieval of valid extents from the stripes, while also avoiding any unnecessary read operations of the bottom-down approach.

Optimized segment cleaning technique

A cluster-wide consistency checker ensures that two file systems of a storage input/output (I/O) stack executing on each node of a cluster are self-consistent as well as consistent with respect to each other. The file systems include a deduplication file system and a host-facing file system that cooperate to provide a layered file system of the storage I/O stack. The deduplication file system is a log-structured file system managed by an extent store layer of the storage I/O stack, whereas the host-facing file system is managed by a volume layer of the stack. Illustratively, each log-structured file system implements a key-value store and cooperates with other nodes of the cluster to provide a cluster-wide (global) key-value store. The consistency checker verifies and/or fixes on-disk structures of the layered file system to ensure its consistency. To that end, the consistency checker may determine whether there are inconsistencies in the key-value store and, if so, reconciles those inconsistencies from a client (volume layer) perspective.

Consistency checker for global de-duplication clustered file system

A system and method for handling un-partitioning of a computer network are disclosed. Routing information may be stored on a first node in the network when the network is partitioned into a first group of nodes and a second group of nodes, where the first node is in the first group of nodes. Storing the routing information on the first node may include storing information usable to route messages to nodes in the first group of nodes. A technique for determining when the second group of nodes has become un-partitioned from the first group of nodes (i.e., for determining when the partition has been repaired) may be employed. In response to determining that the second group of nodes has become un-partitioned from the first group of nodes, the routing information on the first node may be modified. Modifying the routing information on the first node may include storing information useable to route messages to nodes in the second group of nodes.

Handling un-partitioning of a computer network

A set of probabilities may indicate the chance of finding the latest version of a data item in any particular data image layer of a data image. A client may use the set of probabilities to request the data item from one or more layers that have a higher probability of having that version of the data item. A client may query more recently created layers to determine whether they have a version of the data item. If any of the more recently created data image layers has a version of the data item, a client may request the data item from the most recent of the more recently created layers and may discard any data returned from the originally requested layer. A client may update the historical probabilities based on which layers contain versions of the data item.

Probabilistic data locating in sparse data images

A system and method for detecting mirror inconsistencies in a distributed storage environment. Inconsistencies between mirrors may be detected by comparing session tags among individual storage devices. Each data write may include a tag and storage devices may calculate session tags from the write tags. Additionally, a storage device may keep a history including tags from recent writes. When a client sessions ends, a metadata server may compare the respective session tags from different storage devices to determine whether the mirrors are synchronized. If the session tags do not match, the metadata server may examine the metadata histories from the storage devices to determine a cause for the discrepancy in the session tags. If examining the session histories fails to reveal the discrepancy's source, a metadata server may request individual data block checksums from storage devices to attempt a re-synchronization of the mirrored storage devices.

Detecting and repairing inconsistencies in storage mirrors

Various embodiments of a system and method related to a topology and routing model for a peer-to-peer network are disclosed. A plurality of nodes may be coupled to each other to form a peer-to-peer network. A routing table may be created on each node in the peer-to-peer network. Each node may be operable to route messages to other nodes in the peer-to-peer network using information stored in the routing table. Messages may be propagated among nodes in the peer-to-peer network in a decentralized manner. For example, the peer-to-peer network may not utilize centralized servers of any kind. Each node in the peer-to-peer network may perform substantially the same routing functionality.

Topology and routing model for a computer network

A method, system and computer accessible medium for expiring access tokens in preparation for freezing file images. A metadata server may maintain a next scheduled quiesce time and may issue access tokens configured to expire before the next scheduled quiesce time. A metadata server may set an access token's expiration time to a maximum expiration time indicated by the next scheduled quiesce time or may set an access token's expiration time to a default expiration time if the default expiration time is earlier than the maximum expiration time. A storage device may recognize and enforce the access token's expiration time.

Expiration of access tokens for quiescing a distributed system

A method and system for implementing a global name space service. The method may include receiving a file system unique identifier corresponding to a particular file and a human-readable name corresponding to the same file. The method may further include binding the human-readable name to the file system unique identifier, thereby creating a mapping between the human-readable name and the file system unique identifier. The system may include a processor coupled to a memory and to a global name space service manager. The global name space service manager may be configured to receive a file system unique identifier corresponding to a particular file and a human-readable name corresponding to the same file. The global name space service manager may be further configured to bind the human-readable name to the file system unique identifier, thereby creating a mapping between the human-readable name and the file system unique identifier.

Growing community of inventors

Scott Valley, CA, United States of America

John A Muth