NVRAM data organization using self-describing entities for predictable recovery after power-loss

A method comprising:

• executing, by a node coupled to a plurality of solid state drives (SSDs), a storage input/output (I/O) stack having a plurality of layers including a persistence layer for configuring a portion of a non-volatile random access memory (NVRAM) as a log to record in sequence write requests received from a host and to acknowledge successful receipt of the write requests to the host;
• organizing the log as a set of entries, each entry including (i) write data of a write request, (ii) a previous offset referencing a previous entry of the log and (iii) an outstanding entries field indicating a number of outstanding I/O operations in the log at the time that the respective entry is created;
• and
• after a power loss, replaying the log in reverse sequential order using the previous offset of each entry to traverse the log so as to recover the acknowledged write requests, wherein recovery of a last write request received from the host is enabled by the replay of log the logs corresponding to write requests pending in the storage I/O stack at a time of creation of a tail entry of the log.

A non-transitory computer readable medium including program instructions for execution on one or more processors, the program instructions when executed operable to:

• implement a storage input/output (I/O) stack having a plurality of layers including a persistence layer for configuring a portion of a non-volatile random access memory (NVRAM) as a log to record in sequence write requests received from a host and to acknowledge successful receipt of the write requests to the host;
• organize the log as a set of entries, each entry including (i) write data of a write request, (ii) a previous offset referencing a previous entry of the log and (iii) and an outstanding entries field indicating a number of outstanding I/O operations in the log at the time that the respective entry is created;
• and
• after a power loss, replay the log in reverse sequential order using the previous offset of each entry to traverse the log so as to recover the acknowledged write requests, wherein recovery of a last write request received from the host is enabled by the replay of the logs corresponding to write requests pending in the storage I/O stack at a time of creation of the tail entry of the log.