Abstract: Technology is disclosed for QoS (Quality of Service) enforcement with reduced contention. For commands where the QoS wait queue is empty and the QoS credit count is sufficient, an atomic operation subtracts the credits required to process the command from the QoS credit count without locking, and the command is processed. For other commands, credit generation is attempted by first reading a lock flag without obtaining a lock. If the lock flag is clear, an atomic operation sets it, and credit generation is performed if sufficient time has passed since a last credit generation. Credit generation stores a sum of the current QoS credit count and the number of newly generated credits into a local credit count, and processes commands in the QoS wait queue for which sufficient credits exist in the local credit count. When credit generation completes, the local credit count is stored into the QoS credit count.

Abstract: Technology is disclosed for QoS (Quality of Service) enforcement with reduced contention. For commands where the QoS wait queue is empty and the QoS credit count is sufficient, an atomic operation subtracts the credits required to process the command from the QoS credit count without locking, and the command is processed. For other commands, credit generation is attempted by first reading a lock flag without obtaining a lock. If the lock flag is clear, an atomic operation sets it, and credit generation is performed if sufficient time has passed since a last credit generation. Credit generation stores a sum of the current QoS credit count and the number of newly generated credits into a local credit count, and processes commands in the QoS wait queue for which sufficient credits exist in the local credit count. When credit generation completes, the local credit count is stored into the QoS credit count.

Abstract: Deferred memory page allocation commands and non-deferred memory page allocation commands are identified within host I/O commands received by a data storage system. For each one of those received host I/O commands that are identified as a deferred memory page allocation command, QoS (Quality of Service) policy enforcement is performed before any memory pages are allocated by the data storage system to store host data indicated by the received command. Host I/O commands that are identified as deferred memory page allocation commands include read commands, and host I/O commands that are identified as non-deferred memory page allocation commands include in-capsule write commands. For commands identified as deferred memory page allocation commands, enforcing QoS policy before any memory pages are allocated to store host data indicated by the command avoids the possibility of enqueueing the allocated memory pages onto the QoS wait queue, thus conserving data storage system memory resources.

Abstract: A method, computer program product, and computing system for associating a plurality of NVMe IO submission queues with an NVMe IO completion queue. An NVMe IO completion queue polling frequency may be defined for a plurality of threads configured to provide one or more IO requests to the plurality of NVMe IO submission queues. The NVMe IO completion queue may be polled based upon, at least in part, the NVMe IO completion queue polling frequency defined for the plurality of threads.

Abstract: A method, computer program product, and computer system for directing, by a computing device, an incoming Non-Volatile Memory express (NVMe) command to a kernel driver. The kernel driver may enqueue the incoming NVMe command until fetched by a user space. The NVMe command may be fetched from the kernel driver for processing. The NVMe command may be pushed to a user space block device of the user space.

Abstract: A method, computer program product, and computer system for using, by a computing device, free central processing unit (CPU) cycles for processing Non-Volatile Memory express Over Fiber (NVMeoF)/Transmission Control Protocol (TCP) sockets. Using the free CPU cycles for processing the NVMeoF/TCP sockets may include exposing a character device with a callback that iterates over the NVMeoF/TCP sockets for processing.

Abstract: A method, computer program product, and computing system for associating a plurality of NVMe IO submission queues with an NVMe IO completion queue. An NVMe IO completion queue polling frequency may be defined for a plurality of threads configured to provide one or more IO requests to the plurality of NVMe IO submission queues. The NVMe IO completion queue may be polled based upon, at least in part, the NVMe IO completion queue polling frequency defined for the plurality of threads.

Abstract: A method, computer program product, and computer system for directing, by a computing device, an incoming Non-Volatile Memory express (NVMe) command to a kernel driver. The kernel driver may enqueue the incoming NVMe command until fetched by a user space. The NVMe command may be fetched from the kernel driver for processing. The NVMe command may be pushed to a user space block device of the user space.

Abstract: A method, computer program product, and computer system for using, by a computing device, free central processing unit (CPU) cycles for processing Non-Volatile Memory express Over Fiber (NVMeoF)/Transmission Control Protocol (TCP) sockets. Using the free CPU cycles for processing the NVMeoF/TCP sockets may include exposing a character device with a callback that iterates over the NVMeoF/TCP sockets for processing.

Abstract: A system and method for optimized stream management are provided. The method includes retrieving priority data; determining, in real-time, a current priority tree based on the retrieved priority data, wherein the priority tree includes at least one node representing a stream; identifying, based on the retrieved priority data, at least one relative weight of the at least one node; determining an effective weight based on each identified relative weight; and filling a buffer based on the current priority tree and the at least one effective weight.

Abstract: A system and method for optimized stream management are provided. The method includes retrieving priority data; determining, in real-time, a current priority tree based on the retrieved priority data, wherein the priority tree includes at least one node representing a stream; identifying, based on the retrieved priority data, at least one relative weight of the at least one node; determining an effective weight based on each identified relative weight; and filling a buffer based on the current priority tree and the at least one effective weight.