Method and system for implementing an unordered delivery of data between nodes in a clustered storage system

A computer-implemented method comprising:

• sending, by a local node, a mirroring entry to a remote node according to a transmission order, the mirroring entry comprising a write log stored in a local log cache and an order indicator for the write log in the local log cache, the order indicator comprising an address corresponding to a storage location in the local log cache and an address offset to partition write requests according to a respective write source;
• receiving, by the remote node, the mirroring entry across one of a plurality of network connections between the local node and the remote node;
• storing the mirroring entry at the remote node based on an arrival order, the arrival order different from the transmission order;
• storing the write log from the mirroring entry at a location in a remote log cache of the remote node according to the address and address offset of the order indicator to produce a consistent view between the remote log cache and the local log cache partitioned according to the respective write source;
• executing, by the local node, a write operation for the write log on a shared storage accessed by the local node and the remote node;
• transmitting a new mirroring entry comprising an update of the local log cache from the local node to the remote node;
• and
• processing the update on the remote log cache to avoid a duplicate write operation for the write log on the shared storage.

A computer-implemented method comprising:

• generating, by a local node in a storage system cluster (cluster), a write log for a plurality of write requests targeted for shared storage of the cluster and storing the write log in a local log cache of the local node, each of the plurality of write requests having a respective write source;
• generating a mirroring entry comprising the write log and an order indicator of the write log in the local log cache, the order indicator comprising an address corresponding to a storage location in the local log cache and an address offset to partition each write request according to the respective write source;
• transmitting the mirroring entry from the local node to a remote node across one of a plurality of network connections between the local node and the remote node;
• storing the mirroring entry to a next available storage location in an interim cache of the remote node;
• processing, at a later predetermined time, the mirroring entry in the interim cache to produce an ordered write log to a storage location in a remote log cache based on the order indicator and partitioned according to the respective write source;
• executing, by the local node, a write operation for the write log on the shared storage;
• transmitting a new mirroring entry comprising an update of the local log cache from the local node to the remote node;
• and
• processing the update on the remote log cache to avoid a duplicate write operation for the write log on the shared storage.

A computer-implemented method comprising:

• receiving, by a remote node in a storage system cluster (cluster) across a network, a mirroring entry associated with a write log stored at a local node in the cluster, the mirroring entry comprising the write log and an order indicator of the write log, the order indicator comprising an address corresponding to a storage location in the write log and an address offset to partition a plurality of write requests according to respective write sources;
• storing the mirroring entry to a next available storage location in an interim cache of the remote node based on an arrival order of the mirroring entry at the remote node;
• deferring processing of the mirroring entry in the interim cache until a later predetermined time for producing a consistent view between a remote log cache and a local log cache partitioned according to the respective write sources;
• receiving a new mirroring entry comprising an update of the write log from the local node;
• and
• processing the update on the remote log cache to avoid a duplicate write operation for the write log on a shared storage accessed by the local node and the remote node.

A system comprising:

• a processor;
• and
• a memory coupled to the processor and configurable for storing instructions, which, when executed by the processor cause the processor to:
• generate by an entry generator, at a local node of a cluster, a mirroring entry comprising a write log for a plurality of write requests targeted for shared storage of the cluster, each of the plurality of write requests having a respective write source, the mirroring entry further comprising an order indicator of the write log in a local log cache of the local node, the order indicator comprising an address corresponding to a storage location in the local log cache and an address offset to partition each write request according to the respective write source;
• transmit by a mirroring module, based on a transmission order, the mirroring entry from the local node to a remote node across one of a plurality of network connections and store the mirroring entry at the remote node based on an arrival order of the mirroring entry at the remote node;
• process by a deferred processing engine the mirroring entry in a remote log cache of the remote node at a later predetermined time to produce a consistent view between the local log cache and the remote log cache partitioned according to the respective write source;
• the entry generator configured to complete, at the local node, a write operation associated with the write log on the shared storage and transmit a new mirroring entry to the remote node to indicate a completed write operation;
• and
• the mirroring module configured to update the remote log cache with the new mirroring entry to indicate the completed write operation associated with the write log.

A system comprising:

• a processor;
• and
• a memory coupled to the processor and configurable for storing instructions, which, when executed by the processor cause the processor to:
• generate, at a local node in a storage system cluster (cluster), a write log for a plurality of write requests targeted for shared storage of the cluster and storing the write log in a local log cache of the local node, the local log cache implementing an ordering constraint on the write log;
• generate a mirroring entry comprising the write log and an order indicator of the write log in the local log cache, the order indicator comprising an address corresponding to a storage location in the local log cache and an address offset to partition each of the write requests according to a respective write source;
• transmit the mirroring entry from the local node to a remote node across one of a plurality of network connections between the local node and the remote node;
• store the mirroring entry according to an arrival order at the remote node;
• process, at the remote node the mirroring entry according to the address and address offset of the order indicator to produce an ordered write log in a remote log cache partitioned according to the respective write source;
• execute, by the local node, a write operation for the write log on the shared storage;
• transmit a new mirroring entry comprising an update of the local log cache from the local node to the remote node;
• and
• process the update on the remote log cache to avoid a duplicate write operation for the write log on the shared storage.

A system comprising:

• a processor;
• and
• a memory coupled to the processor and configurable for storing instructions, which, when executed by the processor cause the processor to:
• send, at a local node, a mirroring entry to a remote node according to a transmission order, the mirroring entry comprising a write log stored in a local log cache and an order indicator for the write log in the local log cache, the order indicator comprising an address corresponding to a storage location in the local log cache and an address offset to partition write requests according to each respective write source;
• receive, at the remote node, the mirroring entry across one of a plurality of network connections between the local node and the remote node;
• store the mirroring entry at the remote node based on an arrival order, the arrival order different from the transmission order;
• store the write log from the mirroring entry at a location in a remote log cache of the remote node according to the address and address offset of the order indicator to produce a consistent view between the remote log cache and the local log cache partitioned according to the respective write source;
• execute, by the local node, a write operation for the write log on a shared storage accessed by the local node and the remote node;
• transmit a new mirroring entry comprising an update of the local log cache from the local node to the remote node;
• and
• process the update on the remote log cache to avoid a duplicate write operation for the write log on the shared storage.