International Business Machines Corporation

Chengdu Huawei Symantec Technologies Co., Ltd.

Michael Anthony Ko

James T Brady

Damon W Finney

Michael Howard Hartung

Renato John Recio

Jaishankar Moothedath Menon

David R Nowlen

Jacobo A Vargas

Noah R Mendelsohn

Binny Sher Gill

Alan Frederic Benner

Gregory Francis Pfister

Wendy Ann Belluomini

Donald J Lang

Richard Francis Freitas

Michael T Benhase

Windsor Wee Sun Hsu

William Todd Boyd

Douglas James Joseph

John Lewis Hufferd

Shauchi Ong

Lan Huang

Narasimhareddy L Annapareddy

Mircea R Gusat

Prasenjit Sarkar

John Stephen Muhich

Norman Ken Ouchi

Kent R Demke

Paul W Hunter

Jerold D Dwire

Joanne L Mumola

Michael James Rayfield

Shin-Yuan Tzou

Narasimha Lakshmi Annapareddy

Jai M Menon

A system and method for transmitting packets over two different network protocols without protocol conversion in any switches. A computer system comprises host computers and target storage arrays, or targets, coupled to one another through a Enhanced Ethernet network. Each of the host computers and the targets is configured to transmit encapsulated packets, such as a Fiber Channel over Ethernet (FCoE) packet. During system configuration, each of the targets is set to be the only target included in a virtual local area network (VLAN) with a corresponding unique VLAN identifier (ID). A given host computer logins to multiple assigned targets using a predefined Fiber Channel protocol. In response to a login request, a corresponding target assigns and conveys a N_Port ID that comprises a VLAN ID and a unique Host ID to the host computer in a reply message. A virtual link is established between the host computer and the target storage array.

System and method for supporting fibre channel over ethernet communication

Embodiments of the invention are directed to reducing write amplification in a cache with flash memory used as a write cache. An embodiment of the invention includes partitioning at least one flash memory device in the cache into a plurality of logical partitions. Each of the plurality of logical partitions is a logical subdivision of one of the at least one flash memory device and comprises a plurality of memory pages. Data are buffered in a buffer. The data includes data to be cached, and data to be destaged from the cache to a storage subsystem. Data to be cached are written from the buffer to the at least one flash memory device. A processor coupled to the buffer is provided with access to the data written to the at least one flash memory device from the buffer, and a location of the data written to the at least one flash memory device within the plurality of logical partitions. The data written to the at least one flash memory device are destaged from the buffer to the storage subsystem.

Reducing write amplification in a cache with flash memory used as a write cache

According to the present invention, there is provided a method of providing a WORM storage system, the method including a sector-append capability. The method includes receiving data to be written to a WORM storage system. In addition, the method includes identifying a target sector at which the data is to be written. Also, the method includes determining if the received data can be added to the target sector. Moreover, the method includes adding the received data to the target sector if it is determined that the received data can be added to the target sector.

System and method for enabling efficient small writes to worm	 storage

The invention provides a method and apparatus for network congestion management. The method includes inserting a probe frame into data traffic in the network from a first endpoint in the network, reflecting the probe frame from a second endpoint in the network back to the first endpoint, receiving the reflected probe frame at the first endpoint via the network, determining a round-trip delay based on flow of the probe frame through the network, and throttling network traffic according to the determined delay to manage network traffic congestion.

Method and apparatus for end-to-end network congestion management

A system and method for providing a memory region/memory window (MR/MW) access notification on a system area network are provided. Whenever a previously allocated MR/MW is accessed, such as via a remote direct memory access (RDMA) read/write operation, a notification of the access is generated and written to a queue data structure associated with the MR/MW. In one illustrative embodiment, this queue data structure may be a MR/MW event queue (EQ) data structure that is created and used for all consumer processes and all MR/MWs. In other illustrative embodiments, the EQ is associated with a protection domain. In yet another illustrative embodiment, an event record may be posted to an asynchronous event handler in response to the accessing of the MR/MW. In another illustrative embodiment, a previously posted queue element may be used to generate a completion queue element in response to the accessing of the MR/MW.

Providing a memory region or memory window access notification on a system area network

According to one embodiment, a method for using flash memory in a storage cache comprises receiving data to be cached in flash memory of a storage cache, at least some of the received data being received from at least one of a host system and a storage medium, selecting a block of the flash memory for receiving the data, buffering the received data until sufficient data has been received to fill the block, and overwriting existing data in the selected block with the buffered data. According to another embodiment, a method comprises receiving data, at least some of the data being from a host system and/or a storage medium, and sequentially overwriting sequential blocks of the flash memory with the received data. Other devices and methods for working with flash memory in a storage cache according to various embodiments are included and described herein.

Achieving a high throughput in a storage cache application using a flash solid state disk

Mechanisms for providing a memory region/memory window (MR/MW) access notification on a system area network are provided. Whenever a previously allocated MR/MW is accessed, such as via a remote direct memory access (RDMA) read/write operation, a notification of the access is generated and written to a queue data structure associated with the MR/MW. In one illustrative embodiment, this queue data structure may be a MR/MW event queue (EQ) data structure that is created and used for all consumer processes and all MR/MWs. In other illustrative embodiments, the EQ is associated with a protection domain. In yet another illustrative embodiment, an event record may be posted to an asynchronous event handler in response to the accessing of the MR/MW. In another illustrative embodiment, a previously posted queue element may be used to generate a completion queue element in response to the accessing of the MR/MW.

Disclosed are embodiments of a storage area network (SAN), a network interface card and a method of managing data transfers. These embodiments overcome the distance limitation of the Serial Attached Small Computer System Interface (SAS) physical layer so that SAS storage protocol may be used for communication between host systems and storage controllers. Host systems and storage controls are connected via an Ethernet interface (e.g., a legacy Ethernet or enhanced Ethernet for datacenter (EED) fabric). SAS storage protocol is layered over this Ethernet interface, providing commands and transport protocol for information exchange. Since the Ethernet interface has its own physical layer, the SAS physical layer is unnecessary and, thus, so is the SAS distance limitation. If legacy Ethernet is used, over-provisioning is used to avoid packet drops, or alternatively, TCP/IP is supported in order to recover from packet drops. If EED is used, congestion management as well as priority of service functions are provided by the EED protocols.

Layering serial attached small computer system interface (SAS) over ethernet

A memory leveling system updates physical memory blocks, or blocks, to maintain generally even wear. The system maintains an update count for each block, incrementing a wear level count when the update count reaches a wear level threshold. The system compares a wear level of blocks to determine whether to update a block in place or move data on the block to a less-worn physical block. The system groups the blocks into wear level groups identified by a common wear level to identify blocks that are being worn at a faster or slower than average rate. If an empty block count of a least worn group drops below a threshold, the system moves data from one of the blocks in the least worn group to an empty block in a most worn group.

Updating a memory to maintain even wear

A packet classifier within a network adapter detects an application packet header belonging to a packet in a data stream associated with an application that uses a network protocol, such as TCP/IP, SNA or IPX. Offsets included within the application header are identified and a plurality of registers is loaded with the identified offsets. Each of a plurality direct data placement patterns are masked with contents of the loaded registers. Each direct data placement pattern is associated with an application packet header and includes a corresponding I/O context. Additionally, each direct data placement pattern includes a corresponding connection path for direct placement of a payload corresponding to the detected application header in a memory that is utilized by the application.

Method and system for internet transport acceleration without protocol offload

Mechanisms for preventing IP spoofing and facilitating parsing of private data areas in system and network connection requests are provided. With these mechanisms, an identifier, such as the high order bit of a Q_Key, is utilized to determine if a communication connection request originates with a privileged process. A second identifier is used to specify whether a private data area of a communication connection request utilizes predefined fields of a predefined structure or format. Only when the first identifier specifies that the request originates from a privileged process is the processing of the request permitted to be performed. Based on the setting of the second identifier, specific information is retrieved from the predefined fields of the private data area for use in establishing the requested communication connection.

Preventing IP spoofing and facilitating parsing of private data areas in system area network connection requests

Mechanisms for collective send operations on a system area network are provided. The mechanisms of the illustrative embodiments provide for the creation, modification, and removal of collective send queues (CSQs) that allow the upper level protocol (ULP) used by a consumer to send the same message to a collective set of queue pairs (QPs). In order to use the transport services of a CSQ, a consumer process posts a write work request (WR) to the CSQ. The write WR causes a write work queue element (WQE) to be generated and placed in the CSQ. A channel interface (CI) is provided that effectively copies the write WQE to all of the send queues (SQs) of the QPs in the QP set associated with the CSQ. When all the QPs complete processing of their respective write WQEs, the HCA releases all data segments referenced by the write WR.

Collective send operations on a system area network

Provided are a method, system, and program for transferring data between an initiator node and target node. A request is received conforming to a first data transfer protocol at the initiator node to transmit to the target node. A reference to a memory location is obtained to use to transfer the request to the target node. At least one function is called that executes in a user address space of the initiator node, wherein the initiator node includes a kernel address space and the user address space. The at least one function executing in the user address space interfaces with an adaptor to transmit the request and reference to the memory location to the target node using a second data transfer protocol.

Method, system, and program for executing data transfer requests

Provided are a method, system, and program for communicating data transfer requests between data transfer protocols. A request is received conforming to a first data transfer protocol at an initiator node to transmit to a target node, wherein the request includes a request identifier in the first data transfer protocol. A reference is obtained to a memory location to use for the request using a second data transfer protocol, wherein the reference is used by the second data transfer protocol. An entry is added to a map associating the reference to the memory location for the second data transfer protocol with the request identifier for the first data transfer protocol. The second data transfer protocol is used to transfer the request with the request identifier and the reference to the memory location.

Growing community of inventors

San Jose, CA, United States of America

Michael Anthony Ko