摘要:
A method, system, computer system and computer program product to synchronize data and a snapshot of the data taken at a given point in time. Persistent data change maps are used to track changes made to data after a snapshot of the data is taken. Changes to the data are tracked using a persistent accumulator map, and changes to the data with respect to a second set of data are tracked using a persistent volume map. The persistent accumulator map is updated with each update of the data. Persistent volume maps are updated when a subsequent snapshot of the data is taken. Only changes to the data made after the snapshot was taken are applied to synchronize the snapshot with the data so that all of the data is not copied. Snapshots can be located in a physically separate location from the data itself.
摘要:
One goal of consistency interval replication is to achieve a consistent copy of data generated by independent streams of writes from nodes in a clustered/distributed environment. Two writes to the same block from different nodes may arrive at a replication target in a different order from the order in which they were written to primary storage. A consistency interval coordinator may analyze a list of blocks modified during a consistency interval to determine conflict blocks written to by two different nodes during the same consistency interval. Conflict resolution may involve a node reading data for a conflict block from primary storage and forwarding it to the replication target or a node completing a suspended in-progress write for the conflict block. Once the conflicts have been resolved, the replication target may checkpoint the data modified during the interval and nodes may resume writes to the conflict blocks for the new interval.
摘要:
Various embodiments of systems and methods are disclosed for tracking valid regions of a working volume. State information identifies which regions of the working volume are currently valid. When the volume is created, the state information can be initialized to a value that identifies all regions of the volume as being invalid. The invalid regions do not need to be synchronized, since there will not be any need to reconstruct the data within those regions to a particular value. Accordingly, volume initialization, which synchronizes redundant data (e.g., RAID parity or a mirrored copy) with application data in the invalid regions, can be delayed. As the volume is accessed by an application, the redundant data associated with the regions being accessed is synchronized, and the state information is updated to indicate that those regions are valid.
摘要:
Disclosed is a method and apparatus for optimizing memory space and improving the write performance in a data processing system having a data volume with multiple virtual copies thereof. In one embodiment of the method, a first virtual copy of a primary data volume is created. Thereafter, first data of the primary data volume is modified. A second virtual copy of the primary data volume is created after modification of the first data thereof. A write-data transaction for modifying second data of the modified primary data volume is generated after creation of the second virtual copy. The second data of the modified primary data volume is copied to memory allocated to store data of the second virtual copy. The second data of the modified primary data volume is modified after the second data is copied to the memory allocated to store data of the second virtual copy.
摘要:
A method may involve: an application executing on a host initiating a write to a stripe in a storage volume; in response, the host requesting permission from a coordinator to perform the write to a set of data (e.g., a stripe or other group of data from which redundant information is calculated according to a RAID protocol) in a storage volume; prior to receiving permission to write to the storage volume from the coordinator, the host initiating storage of write data associated with the write to an associated one of a plurality of logs; the host waiting before performing the write to the storage volume; and, if one or more additional writes targeting the same set of data (e.g., the same stripe) are initiated while waiting, the host combining redundant information (e.g., parity) calculations for the writes into a single redundant information calculation.
摘要:
Disclosed is a method and apparatus for refreshing a copy of a data volume. In one embodiment of the method first and second data portions of a data volume are copied to first and second memory blocks, respectively, of a memory coupled to a computer system. First and second bits of a first map stored in memory are then set, wherein the first and second bits correspond to the first and second memory blocks, respectively. The first data portion of the data volume is modified after the first data portion is copied to the first memory block. A first bit in a second map stored in memory is set after data of the first data portion is modified. An instruction is generated to refresh the data contents of the first and second memory blocks. The first bit of the first map is cleared in response to generation of the refresh instruction.
摘要:
A virtual copy of data stored in a first memory is created in a second memory. Creating the virtual copy includes, in one embodiment, creating first and second tables in memory each one of which comprises a plurality of multibit entries. Each entry of the first table corresponds to a respective memory region of the first memory. Each entry of the second table corresponds to a respective memory region of the second memory. The first bit of the first and second tables indicates whether the corresponding memory region of the first and second memories, respectively, contains valid data. The second bit of the first and second tables indicates whether data in the corresponding memory region of the first and second memories, respectively, has been modified since the creation of the first and second tables, respectively.
摘要:
A log manager may store a validation indicator with each data block of a log entry. The log manager may create a header block for each log entry that may include a validation indicator or tag. Such a validation indicator may be stored within the metadata for each data block. The validation indicator may additionally be stored in the metadata for the header block and it may be stored in header block itself. When recovering such a log, according to some embodiments, the validation indicators for each data block are checked against the validation indicator in the header block.
摘要:
Disclosed is an apparatus and method for transforming unrelated data volumes into related data volumes. The present invention is employed after creation of first and second unrelated data volumes. In one embodiment, the second data volume is refreshed to the data contents of the first data volume so that the second data becomes a PIT copy of the first data volume. Refreshing the second data volume includes overwriting all data of the second data volume with data copied from the first data volume. However, before all data of the second data volume is overwritten with data copied from the first data volume, data of the first data volume can be modified.
摘要:
Disclosed is an apparatus or method performed by a computer system for creating a hierarchy of data volumes. Each data volume in the hierarchy is a point-in-time (PIT) copy of another data volume in the hierarchy or a PIT copy of a data volume V. In one embodiment of the apparatus or method, the contents of a first data volume in the hierarchy can be refreshed to the contents of a second data volume in the hierarchy such that the first data volume becomes a PIT copy of the second data volume. Before the first data volume is fully refreshed to the contents of the second data volume, data of the first data volume can be read or modified.