Write-once-read-many in cloud for file systems backed up to object storage

A non-transitory machine-readable medium, comprising executable instructions that, when executed by at least one processor, facilitate performance of operations, comprising:

• determining changes to file data stored in a file system relative to an initial version of an object chunk representative of the file data in an object store, wherein the file data is determined to comprise a defined write-once-read-many data retention policy, wherein the determining changes to file data comprises a determination that the initial version of the object chunk satisfies a criterion of the defined write-once-read-many data retention policy;
• storing, by utilizing a pseudo-copy-on-write replication of the file system to the object store and based on the determining changes, changed data, of the file data that is copied, from the file system to an updated version of the object chunk;
• storing, via a transfer component and based on the determining changes, unchanged data;
• of the file data that is copied, from the initial version of the object chunk to the updated version of the object chunk;
• generating, based on the copied changed data and the copied unchanged data, a snapshot object, wherein the snapshot object comprises an empty object that comprises metadata pointing from the updated version of the object chunk to the initial version of the object chunk;
• designating the updated version of the object chunk as a head object in the object store, wherein the head object stores, in the object store, all file data for a file represented by the object chunk;
• maintaining the head object as a fully populated repository of the file data, wherein copy-on-write redirect objects, which are associated with respective dataset versions of the file system, comprise empty objects that store metadata pointers to the head object and do not contain file data, and wherein the defined write-once-read-many data retention policy for each respective dataset version is satisfied through retention of the head object, without storing duplicate file data for each respective dataset version of the respective data set versions in the object store;
• and
• in response to copying the changed data and copying the unchanged data,
• changing the head object from the initial version of the object chunk to the updated version of the object chunk.

A method, comprising:

• detecting, by a system comprising at least one processor, modifications to a file of a file system relative to a first version of an object chunk, representative of a portion of the file, in an object store, wherein the file is determined to comprise a defined data retention rule, wherein the detecting modifications to a file comprises a determination that the first version of the object chunk satisfies a criterion of the defined data retention rule;
• wherein the object chunk is identified using a dataset identifier applicable to the file system and a file identifier applicable to the file,
• wherein the file identifier is allocated as a globally unique identifier for the file,
• wherein the file identifier remains constant across all versions of the file,
• wherein the dataset identifier is allocated as a globally unique identifier for each dataset version,
• wherein the dataset identifier changes with each incremental replication while the file identifier remains constant with each incremental replication, and
• wherein a tuple comprising the file identifier and the dataset identifier is utilized to identify respective concrete versions of the file in the object store;
• copying, by the system and based on the modifications to a file, modified data of the file from the file system to a second version of the object chunk in the object store;
• copying, by the system and based on the modifications to a file, unmodified data of the file from the first version of the object chunk to the second version of the object chunk in the object store;
• generating based on the copying modified data and the copying unmodified data, by the system, a snapshot object, wherein the snapshot object comprises an empty object that comprises metadata pointing from the second version of the object chunk to the first version of the object chunk;
• designating, by the system, the second version of the object chunk as a head object in the object store,
• wherein respective object names in the object store are hierarchically rendered using respective dataset identifiers and respective file identifiers, and
• wherein the tuple is encoded into an object name of the object chunk as a hierarchical path that comprises a dataset-identifier component and a file-identifier component, and wherein object names in the object store are organized in a hierarchy according to respective dataset identifiers and respective file identifiers;
• and
• in response to the copying modified data and the copying unmodified data,
• changing the head object from the first version of the object chunk to the second version of the object chunk.