Infinidat Ltd.

Yechiel Yochai

Haim Kopylovitz

Leo Corry

Efraim Zeidner

Ido Ben-Tsion

Josef Ezra

Julian Satran

Sivan Tal

Michael Dorfman

Israel Gold

Mike Dorfman

Nir Schenkler

Ido Ben Tsion

Haim Bitner

Ayelet Shani

Joseph Ezra

Amir Frankel

Dan Alon

A method for managing a cache memory of a storage system, the method may include receiving, by a controller of the storage system, an access request related to a data unit; wherein the receiving occurs while (a) the cache memory stores a group of oldest cached data units, and (b) the data unit is stored in a memory module of the storage system the differs from the cache memory; determining, by the controller, a caching category of the data unit; and preventing from caching the data unit in the cache memory when a hit score of the caching category of the data unit is lower than a hit score of the group of oldest cached data units; and caching the data unit in the cache memory when the hit score of the caching category of the data unit is higher than the hit score of the group of oldest cached data units; wherein the hit score of the caching category of the data unit is indicative of a probability of a cache hit per data unit of the caching category.

Self-tuning cache

A method for managing write requests in a storage system, the method may include receiving, by a receiving control node out of a group of control nodes of the storage system, a write request for writing a data unit to the storage system; storing the data unit in a first layer storage that is shared among the control nodes of the group, by the receiving control node; retrieving the data unit from the first layer storage and destaging the data unit to a second layer storage by a destaging control node that belongs to the group and is responsible for destaging the data unit; wherein at a time of the destaging of the data unit to the second layer storage another control node of the group is responsible for destaging another data unit to the second layer storage; wherein the storing of the data unit in the first layer storage is faster than the destaging of the data unit to the second layer storage; and when the destaging control node differs from the receiving control node then each one of the receiving and the storing is not preceded by notifying the destaging control node about the receiving and the storing, respectively.

Shared layered physical space

A method for synchronizing between a primary control node of a storage system and a secondary control node of the storage system, the method may include performing a change in a key-value store that is stored in shared memory of the storage system and includes control metadata for controlling access to data stored in the storage system; the shared memory is accessible to the primary control node and to a secondary control node; generating a change indication that (i) is indicative of the change in the key-value store and (ii) requests the secondary control node to introduce changes in one or more additional data structures that are not stored in the shared memory; the change in the key-value store comprises replacing a previous given value with a current given value that is associated with a given key; sending the change indication to the secondary control node; receiving an acknowledgement from the secondary control node; wherein the acknowledgement indicates at least one of the following: (i) the secondary control node received the change indication and (ii) the secondary control node introduced the changes in the one or more additional data structures that are not stored in the shared memory; and deleting, by the primary control node, the previous given value from the shared memory.

Synchronizing control nodes and a recovery from a failure of a primary control node of a storage system

A method for hash-based writing, the method comprises: receiving a received data entity to be stored in a storage system, wherein the received data entity is associated with received data entity metadata; selecting a selected data structure out of a set of data structures that comprises K data structures; wherein K is a positive integer; wherein for each value of a variable k that ranges between 2 and K, stored data entity metadata that is stored in a k'th data structure out of the set collided with stored data entity metadata that is stored in each one of a first till (k&#x2212;1)'th data structures of the set; calculating an index by applying, on the received data entity metadata, a hash function that is associated with the selected data structure; determining whether an entry that is associated with the index and belongs to the selected data structure is empty; writing to the entry, if the entry is empty, the received data entity metadata, and storing the received data entity in the storage system in response to a location of the entry in the set; and selecting, if the entry is not empty, a new data structure of the set and repeating at least the stages of calculating and determining.

Hash based read and write operations in a storage system

A method for de-duplication, the method may include receiving a request to store in a storage system a received data entity; obtaining a received data entity signature that is responsive to the received data entity; selecting a selected data structure out of a set of data structures that comprises K data structures; wherein K is a positive integer; wherein for each value of a variable k that ranges between 2 and K, a stored data entity signature that is stored in a k'th data structure out of the set collided with stored data entity signatures that are stored in each one of a first till (k&#x2212;1)'th data structures of the set; calculating an index by applying, on the received data entity signature, a hash function that is associated with the selected data structure; determining whether an entry that is associated with the index and belongs to the selected data structure is empty; writing to the entry, if the entry is empty, the received data entity signature, and storing the received data entity in the storage system in response to a location of the entry in the set; selecting, if (a) the entry is not empty and (b) the received data entity signature differs from a stored data entity signature that is stored in the entry, a new data structure of the set, and repeating at least the stages of calculating and determining.

Hash based de-duplication in a storage system

A method that may include determining, for each user of a group of users, a time difference between an event of a first type that is related to a storage of a user data unit of the user within a cache of a storage system and to an eviction of the user data unit from the cache, in response to (a) a service-level agreement (SLA) associated with the user and to (b) multiple data hit ratios associated with multiple different values of a time difference between events of the first type and evictions, from the cache, of multiple user data units of the user; and evicting from the cache, based upon the determination, one or more user data units associated with one or more users of the group.

Cache eviction according to data hit ratio and service level agreement

A method for writing data objects, the method may include accumulating, in a first memory module, multiple new data entities in one or more dirty pages of a data layer; wherein each new data entity and a corresponding older data entity are associated with a same application object; wherein the accumulating comprises storing each new data entity in a page that differs from a page that stores the corresponding older data entity; calculating multiple new sets of descriptors by generating to each new data entity, a new set of descriptors; wherein each set of descriptors comprises descriptors that belong to multiple descriptors layers; wherein the multiple descriptors layers and the data layer belong to an hierarchical data structure; accumulating the multiple new sets of descriptors in one or more dirty pages of one or more descriptors layers; wherein each corresponding older data entity is associated with a corresponding set of descriptors; wherein the accumulating comprises storing each new set of descriptor in a page that differs from a page that stores a corresponding set of descriptors; and writing the multiple new data entities and the multiple new sets of descriptors to a second memory module.

Writing pages to a storage system

There is provided a storage arrangement and a method of operating thereof. The storage arrangement comprises a first storage system and one or more second storage systems operatively coupled to the first storage system. First control layer is operable to handle a first logical address space comprising a first logical group characterized by a plurality of logical block addresses; first control layer comprises a first mapping module handling a first mapping structure associated with first logical group. Each second control layer comprises, respectively, a second mapping module handling a second mapping structure associated with first logical group. The first mapping structure is configured to provide mapping between logical addresses related to first logical group and corresponding addresses related to first physical address spaces, and/or to point to respective second mapping structure configured to provide mapping between these logical addresses and corresponding addresses related to respective second physical address spaces.

Storage arrangement and method using a plurality of storage spaces which have separate control layers and separate mapping structures

A method for generating remote replicates of a logical volume at a remote site, the method comprises performing a plurality of replication cycles for generating a plurality (N) of points in time replicates of the logical volume at the remote site. Wherein for each positive integer n that ranges between 2 and N, a performing of a n'th replication cycle comprises: calculating in parallel and by multiple computerized resources of a storage system, multiple n'th point in time differences of multiple subsets of the logical volume; wherein n'th point in time differences of different subsets are calculated by different computerized resources; wherein a n'th point in time difference of a subset is a difference between an (n&#x2212;1)'th point in time content of the subset and an n'th point in time content of the subset; and sending, in parallel, the multiple n'th point in time differences to the remote site.

High speed asynchronous remote mirroring

A method for generating transaction aware snapshots, the method may include: (a) starting a creation of a certain point in time snapshot for a consistency group that comprises at least one logical volume; (b) detecting that at the certain point in time there is an uncompleted transaction for writing multiple information units to the consistency group; the uncompleted transaction started at a transaction start point in time that precedes the certain point in time; (c) preventing an access to the certain point in time snapshot until a completion of the uncompleted transaction; and (d) associating the multiple information units with at least the certain point in time snapshot.

System, method and a non-transitory computer readable medium for transaction aware snapshot

A method for disk failure protection, the method may include calculating a first set of parity units by processing a first group of sets of data units that are cached in a cache memory of a storage system; calculating a second set of parity units by processing the first group of sets of data units; wherein the calculating of the second set of parity units is responsive to a first shift that was virtually introduced between each set of data units of the first group of sets of data units; and destaging the first group of sets of data units and the first and second sets of parity units to the first group of disks.

Scalable protection scheme for protecting destaged data units

There is provided a storage system comprising a control layer operable to manage a snapshot family (SF) constituted by SF members having hierarchical relations represented by a SF relationship data structure and a method of operating thereof. Upon receiving a command for restoring a writable SF member from a read-only snapshot thereof, the control layer provides a foreground atomic operation. The atomic operation comprises generating a new writable snapshot of the read-only snapshot (a destination SF member), modifying the SF relationship data structure, and re-assigning all logical unit numbers (LUNs) from the writable SF member to be restored to the destination SF member. Upon successful completion of the atomic operation, the control layer sends acknowledgement indicative of successful restoring the writable SF member. The control layer further provides a background process comprising modifying, in correspondence with the provided atomic operation, a SF mapping data structure comprising mappings between logical and physical addresses corresponding to SF members.

Storage system capable of managing a plurality of snapshot families and method of operating thereof

There is provided a storage system capable to maintain a snapshot family comprising a plurality of members having hierarchical relations therebetween, and a method of operating thereof. The method comprises generating a mapping data structure with mapping entries each comprising mappings for mapping a logical address range associated with the mapping entry into physical address ranges respectively correlated to representative members of the snapshot family; omitting mappings corresponding to omitted member(s) different from any of the one or more representative members sharing a respective range of physical addresses; responsive to an access request directed to a certain logical address range and a certain omitted member, determining, using a predefined rule related to the hierarchical and sequential relations between members of the snapshot family, a representative member from the one or more representative members; and responding to the access request by using a physical address range correlated to the representative member.

Storage system and methods of mapping addresses of snapshot families

There is provided a storage system comprising a control layer operable to manage a plurality of snapshot families, each family constituted by snapshot family members having hierarchical relations therebetween. The method of operating the storage system comprises searching a cache memory for an addressed data block corresponding to an addressed LBA and associated with an addressed snapshot family and an addressed SF member. In a case of missing the addressed data block in the cache memory, identifying in the cache memory all dirty data blocks corresponding to the addressed LBA and associated with the addressed snapshot family (candidate data blocks); identifying among SF members of the addressed snapshot family a destage owner, and in a case when the identified destage owner is associated with a candidate data block, reading from the cache memory such candidate data block.

There is provided a storage system and a method of identifying delta-data therein between two points-in-time. The method comprises: generating successive snapshots Sand Scorresponding to the two points-in-time; upon generating the snapshot S, searching the cache memory for data blocks associated with snap_version=i, thereby yielding cached delta-metadata; searching the SF mapping data structure for destaged data blocks associated with snap_version=i, thereby yielding destaged delta-metadata; and joining the cached delta-metadata and the destaged delta-metadata, thereby yielding delta-metadata indicative of the delta-data between points-in-time corresponding to the successive snapshots with Snap_ID=i and Snap_ID=i+1. The method operates with no need in copying, destaging to a dedicated location and/or any other special handling of data that needs to be transmitted to the RS system.

Growing community of inventors

Moshav Aviel, Israel

Yechiel Yochai