Samsung Electronics Co., Ltd.

Venkataratnam Nimmagadda

Muthazhagan Balasubramani

Sandeep Kumar Ananthapalli

Anil Desmal Solanki

Chirag Chinmay

Srinivasa Raju Nadakuditi

Prashant Vishwanath Mahendrakar

Vulligadla Amaresh

Raphel David Johnson

Shruthi Muthukumaran

Dhananjayan A

Renuka Vijay Sapkal

A method of accessing a Non-Volatile Memory express over Fabrics (NVMeoF) memory region of a remote target device includes establishing a NVMeoF transport connection with the remote target device and creating an admin queue by sending a NVMeoF connect command to the remote target device. The method includes accessing the NVMeoF memory region of the remote target device by sending PropertyGet/PropertySet commands to an admin submission queue of the remote target device, performing one of (a) receiving PropertyGet/PropertySet completion queue entries from the remote target device when accessing the NVMeoF memory region of the remote target device is completed, or (b) receiving at least one of RDMA write and RDMA read requests from the remote target device in response to PropertyGet/PropertySet submission queue entries, and receiving completion queue entries from the remote target device when the accessing of the NVMeoF memory region of the remote target device is completed.

METHOD of accessing NVMeoF memory region of a remote target device by a host cpu

A system manages communication between a non-volatile memory express-over fabric (NVMe-oF) host unit and multiple non-volatile memory express-solid state drive (NVMe-SSD) storage devices via a bridge unit. The bridge unit may include sub-modules to control operations. The bridge unit may generate a virtual data memory address corresponding to a scattered gathered list address. The bridge unit may not require a data buffer to store intermediate data. The system may be configured to initiate a memory WRITE/READ transaction to access a virtual data memory corresponding to a physical memory in the bridge unit for performing a data WRITE/READ operation by an NVMe-SSD storage device.

Systems and methods for managing communication between NVMe-SSD storage device(s) and NVMe-of host unit

A first host system in a multipath storage system acts on behalf of a second host system when the second host system is unable to access a solid state drive (SSD). The first host system configures a bitmap table using an SSD controller of the first host system to access memory of the SSD associated with the second host system. The memory accessed on behalf of the second host system may be in a region including persistent memory, base address register memory and/or controller memory buffer of the second host system.

Method and apparatus for accessing at least one memory region of SSD during failover situation in multipath system

A method of accessing a dual line solid-state drive (SSD) device through a network interface and a PCIe EP simultaneously. The method includes: (1) establishing, by the dual line SSD device, a connection with a remote server through the network interface, (2) establishing, by the remote server, an administrative queue with the dual line SSD device, (3) establishing, by the remote server, an input/output queue with the dual line SSD device by posting a command in the administrative queue over the network interface to initiate transfer of data, (4) establishing, by the dual line SSD device, a connection with a local server over the PCIe EP, (5) establishing, by the local server, the administrative queue over the PCIe EP, and (6) establishing, by the local server, the input/output queue by posting the command in the administrative queue over the PCIe EP to initiate transfer of the data.

Method of accessing a dual line SSD device through PCIe EP and network interface simultaneously

A non-volatile memory express over fabrics (NVMeoF) redundant array of independent disks (RAID) controller includes an NVMeoF RAID target module, an NVMeoF RAID implementation module, and an NVMeoF RAID initiator module. The NVMeoF RAID target module receives one or more NVMeoF commands from one or more computer hosts. The NVMeoF RAID implementation module receives the one or more NVMeoF commands from the NVMeoF RAID target module and performs RAID functionalities on the one or more NVMeoF commands. The NVMeoF RAID initiator module receives the one or more NVMeoF commands from the NVMeoF RAID implementation module and transmits the one or more NVMeoF commands to the one or more storage target devices to establish one or more virtual queue pair connections and enable direct data transfer between the one or more computer hosts and the one or more storage target devices.

Method of NVMe over fabric RAID implementation for read command execution

Systems and methods for managing transfer of NVMeoF commands/responses between a host and a target are described. The systems and methods may initiate and convert at least one Input/Output request into at least one Non-Volatile Memory Express over Fabric (NVMeoF) command to access a storage device attached with the target device. A host may transmit the at least one NVMeoF command in a burst mode using a Remote Direct Memory Access (RDMA) Write packet to a pre-registered memory region of the target device. In response to reception of the at least one NVMeoF command, the target device may post at least one NVMeoF completion response corresponding to the at least one NVMeoF command using the RDMA Write packet to a pre-registered memory region of the host.

Non-volatile memory express over fabric messages between a host and a target using a burst mode

Embodiments herein provide a method for managing a read buffer queue in a data storage system to reduce the read buffer size requirements. The method includes receiving a read command from a NVMe host to fetch data from a flash of a NVMe device. Further, the method includes placing a portion of the data fetched from a flash in a read buffer queue in the NVMe device controller. The read buffer queue includes a plurality of segments to accommodate individual portion of the data fetched from the flash. Furthermore, the method includes indicating presence of each portion of data in the plurality of segments of the read buffer queue to the NVMe host.

Method for reducing read buffer size requirements in NVMe based solid state drives

Methods and systems for processing Physical Region Pages (PRP)/Scatter Gather Lists (SGL) entries include splitting a command to be processed into a plurality of sub-commands, storing said plurality of sub-commands in a first set of buffers among a plurality of buffers, processing said plurality of sub-commands from said first set of buffers, storing at least one sub-command that remains after storing the first set of buffers in a second set of buffers, while said plurality of sub-commands in the first set of buffers is being processed and processing said at least one sub-command from said second set of buffers, after processing sub-commands from said first set of buffers.

Methods and systems for processing PRP/SGL entries

A solid state storage device including a non-volatile memory Express (NVMe) controller and configured to manage a Host Memory Buffer (HMB) in a host may be provided. The NVMe controller may be configured to fetch HMB descriptor entries as part of a feature command from the host, partition the HMB logically into a control buffer partition and a data buffer partition, store the HMB descriptor entries pointing to the control buffer partition within the solid state storage device as an HMB pointer list table, and write-back the HMB descriptor entries pointing to the data buffer partition into the control buffer partition of HMB.

Method and system for managing host memory buffer of host using non-volatile memory express (NVME) controller in solid state storage device

A data storage system includes a host having a write buffer, a memory region, a submission queue and a driver therein. The driver is configured to: (i) transfer data from the write buffer to the memory region in response to a write command, (ii) generate a write command completion notice; and (iii) send at least an address of the data in the memory region to the submission queue. The host may also be configured to transfer the address to a storage device external to the host, and the storage device may use the address during an operation to transfer the data in the memory region to the storage device.

Growing community of inventors

Karnataka, India

Venkataratnam Nimmagadda