Emc IP Holding Company LLC

Emc Corporation

Dell Products Inc.

Bhaskar Bora

Mark J Halstead

Benjamin W Yoder

Deepak Vokaliga

Jeffrey Wilson

William R Stronge

Toufic Tannous

Ian Adams

Arieh Don

Michael Ferrari

Sandeep Chandrashekhara

Michael E Specht

Benoit Joseph Merlet

Richard Goodwill

Dharmesh Thakkar

Gaurav Mukul Bhatnagar

Snapshots from a first LSU (R1) on a first storage system (A1) may be replicated to a second replica LSU (R2) on a second storage system (A2), for example, concurrently to remotely replicating (e.g., synchronously) write operations for R1 to R2. A process, P, on A1 executing the replication of the snapshots from R1 to R2 may be a separate process than the one or more processes on A1 executing remote replication of write operations for R1 to R2. During a consistency window on A1, outstanding write operations for R1 at the time the consistency window opened may be logged, and a pair of snapshots, SS1and SS1may be activated on R1 and R2, respectively. After the consistency window has closed, the SS1snapshot metadata and snapshot data may be updated based on the outstanding write operations.

Synchronous remote replication of snapshots

Snapshots may be used to implement remote replication, for example, asynchronously, between a first storage system, A1, and a second storage system, A2. A1 may take a first snapshot, SS2, of a logical storage unit, R1. After the snapshot SS2it taken, any dependent write operations associated with SS2may be reconciled, and differences between SS2and the last snapshot for R1, SS1, may be determined and recorded as a difference set. One or more replication instructions for R1 that include the write operations (or data and metadata corresponding thereto) of the difference set may be sent from A1 to A2. A2 may apply the differences to R2, and then take (activate) a snapshot of R2, SS2, which is a replica of SS2. After A2 activates SS2, A2 may send an acknowledgement to A1 indicating that SS2has been activated, and A2 may take a next snapshot of R1.

Snapshot-based remote replication

Snapshots may be remotely replicated asynchronously from a first LSU (R) on a first storage system (A) to a second replica LSU (R) on a second storage system (A). The storage system Amay open a consistency window to suspend initiating processing of new write operations received on A. While the consistency window is open, Amay: take a first snapshot, SS, of R; record, in association with the first replication cycle, an indication to replicate SSon A; and initiate a next replication cycle to record write operations of the next new write requests to be received from hosts. After initiating a next replication cycle, Amay close the consistency and transmit the first replication cycle to A. Amay apply the write operations of the first replication cycle to R, and then take a second snapshot SSof R, which should be a replica of SS.

Asynchronous remote replication of snapshots

Snapshots are processed without holding all write operations while the snapshots are being activated. Rather than holding all write operations until snapshots are activated, write operations may be allowed to proceed. Snapshot write processing may be temporarily suspended while the snapshots are being activated, including snapshot metadata being updated, while write operations received while the snapshots are being activated are logged. After snapshots have been activated for all logical LSUs for which snapshots were instructed to be activated, the logging of write operations may be stopped, and the logged write entries processed to determine whether any of the logged write operations require updating snapshot information of any logical storage elements (LSEs) of the LSUs. While the logged write operations are being processed, any write operations received from a host for an LSE having a logged write operation may be held until the held operation, or all held operations are processed.

Reducing write delays while taking snapshots

Cache slots on a storage system may be shared between entities processing write operations for logical storage unit (LSU) tracks and entities performing remote replication for write operations for the LSU tracks. If a new write operation is received on a first storage system (S1) for a track of an LSU (R1) when the cache slot mapped to the R1 track is locked by a process currently transmitting data of the cache slot to a second storage system (S2), a new cache slot may be allocated to the R1 track, the data of the original cache slot copied to the new cache slot, and the new write operation for the R1 track initiated on S1 using the new cache slot; while the data of the original cache slot is independently, and perhaps concurrently, transmitted to S2 to be replicated in R2, the LSU on S2 that is paired with R1.

Resolving cache slot locking conflicts for remote replication

Snapshots of a first logical storage unit (LSU) (R1) on a first storage system (A1) may be taken while replication is inactive between A1 and a second storage system (A2), such that these outstanding snapshots are not replicated to a second replica LSU (R2) on A2. Upon replication becoming active, the outstanding snapshots may be replicated to R2 without disrupting or impairing the remote replication of write operations between R1 and R2 as part of standard remote replication. A process on A1 executing the replication of the outstanding snapshots from R1 to R2 may be a separate process than one or more processes on A1 executing standard remote replication including the replication of write operations from R1 to R2. The process may be given low priority on A1 so as to not impair performance of other operations, including standard remote replication and replicating outstanding write operations on A1.

Remote replication of snapshots taken while replication was inactive

Techniques for determining replication modes may include: issuing, while unsolicited data transfer mode is enabled for a first link, first write operations over the first link; issuing, while unsolicited data transfer mode is disabled for the first link, second write operations over the first link; determining a first performance metric for the first link in accordance with the first write operations; determining a second performance metric for the first link in accordance with the second write operations; and determining, in accordance with the first and second performance metrics whether to enable or disable unsolicited data transfer mode for the first link for a first time period. The first and second performance metrics may be response times. Unsolicited data transfer mode, when enabled, allows write data payload to be sent to a target without receiving an acknowledgement regarding receipt of a write command block for the write operation from the target.

Data replication techniques

Maintaining consistency for asynchronous replication of LUNs of a local storage system to corresponding LUNs of a remote storage system includes copying application data written to LUNs to cache slots, copying metadata of LUNs to cache slots, associating a first sequence number to all metadata modifications and all application data writes that are begun after a first time and before a second time, associating a second sequence number, different from the first sequence number, to all metadata modifications and all application data writes that are begun after the second time, and, after completion of all metadata modifications and all application data writes associated with the first sequence number, initiating transfer to the remote storage system of all metadata modifications and all application data writes associated with the first sequence number. Each cache slot may include a field that indicates a corresponding sequence number.

Maintaining consistency of asynchronous replication

A target-less point in time image (snapshot) of a storage volume is allowed to be built after activation, by enabling the snapshot data to be modified to create a crash-consistent replica of the source data after the snapshot has been activated. The data of the snapshot remains immutable from a user standpoint, but the snapshot is able to be quickly activated before all of the data of the snapshot has been included in the snapshot, to thus reduce an amount of time IO operations on the source volume are quiesced. A first snapshot of a storage volume is created on a primary storage system and a corresponding second snapshot of the storage volume is activated on a backup storage system before all the data of the first snapshot is received at the backup storage system. Entries of the activated second snapshot are then changed to point to correct back-end allocations.

Modifying data of activated target-less point-in-time images of storage volumes

In a data storage system in which a first storage array and a second storage array maintain first and second replicas of a production volume, the first storage array is responsive to a write command from a host to send a notification to the second storage array indicating that the replicated production volume will be updated. The notification has information that enables the second storage array to implement pre-processing steps to prepare for subsequent receipt of data associated with the write command. Both storage arrays implement the pre-processing steps at least partly concurrently. When the data associated with the write command is subsequently received, the first storage array writes the data to cache and then sends a copy of the data to the second storage array, i.e. in series. The second storage array then writes the data to cache. Elapsed time between receipt of the write command and returning an acknowledgment to the host may be improved by concurrent pre-processing.

Concurrent remote IO processing for synchronous replication

Architectures and techniques are described that can address challenges associated with engaging a procedure that has the potential to produce beneficial effects in some case and detrimental effects in other cases. One example can be engaging an early IO (EIO) protocol in a synchronous remote data facility (SRDF), but only when doing so will have a positive impact on performance and/or response times.

Early IO control techniques for a remote data facility

Techniques for processing I/O operations may include: receiving, at a first data storage system from a host, a write operation that writes data to a logical device; sending the write operation in a synchronous manner to a plurality of other data storage systems, wherein the first data storage system and the plurality of other data storage systems form a linear chain over which the write operation is transmitted; and sending an acknowledgement to the host regarding completion of the write operation only after the first data storage system and each of the plurality of other data storage systems have acknowledged completion of the write operation.

Described are data replication techniques. Y cycles of writes directed to a first storage device of a first data storage system are collected, Y&#x3e;2. Each of the Y cycles denotes writes directed to the first storage device at an occurrence of a fixed time interval. Writes of cycle N&#x2212;1 directed to the first storage device are transmitted from the first data storage system to a second data storage system. Writes of cycle N&#x2212;2 are applied to a second storage device. An acknowledgement regarding cycle N&#x2212;1 is sent from the second data storage system to the first data storage system responsive to determining that the writes of cycle N&#x2212;1 directed to the first storage device have been received by the second data storage system and that the writes of cycle N&#x2212;2 directed to the first storage device have been applied to the second storage device.

Data transfer techniques with data replication

A method of IO workload selection and prioritization in data mirroring environments includes setting a first weight, a second weight, and a third weight to be used during an iteration of IO processing. The method further includes processing IOs of a first class of IO operations while there are IOs of the first class of IO operations to be processed and while a first weight for the first class of IO operations is positive, processing IOs of a second class of IO operations while there are IOs of the second class of IO operations to be processed and while a second weight for the second class of IO operations is positive, processing IOs of a third class of IO operations while there are IOs of the third class of IO operations to be processed and while a third weight for the third class of IO operations is positive.

IO workload selection and prioritization in data mirroring environments

Described are data replication techniques. Writes of cycle N directed to a first storage device of a first data storage system are received. Writes of cycle N&#x2212;1 directed to the first storage device are transmitted to a second data storage system. Writes of cycle N&#x2212;2 directed to the first storage device are applied to a second storage device of the second data storage system. An acknowledgement regarding cycle N&#x2212;1 is sent to the first data storage system responsive to determining that the writes of cycle N&#x2212;1 directed to the first storage device have been received by the second data storage system and that the writes of cycle N&#x2212;2 have been applied to the second storage device. At least some writes of cycle N directed to the first storage device are transmitted to the second data storage system prior to the first data storage system receiving the acknowledgement regarding cycle N&#x2212;1.

Growing community of inventors

Shrewsbury, MA, United States of America

Bhaskar Bora