摘要:
A method, and corresponding system, is provided for generating replicas or copies of data objects that are used to serve data streams. The method includes determining one of the storage devices is operating at a load above a predetermined overload value. The method continues with identifying the data objects being served by data streams from the near overload storage device and then, selecting one of these served data objects for replication. To create a replica, the method includes writing data from the selected data object to another less busy one of the data storage devices, with the writing being done using data from one or more of the streams being served from the selected data object on the near-overload storage device. During replica creation, one or more streams from the selected data object provide data concurrently to a stream consumer or client device and to the replica.
摘要:
Various embodiments of systems and methods for performing accumulation operations on block operands are disclosed. In one embodiment, an apparatus may include a memory, a functional unit that performs an operation on block operands, and a cache accumulator. The cache accumulator is configured to provide a block operand to the functional unit and to store the block result generated by the functional unit. The cache accumulator is configured to provide the block operand to the functional unit in response to an instruction that uses an address in the memory to identify the block operand. Thus, the cache accumulator behaves as both a cache and an accumulator.
摘要:
Methods and systems for data storage are described herein. In one aspect of the invention, an exemplary process includes receiving a first data being directed to a first storage volume, receiving a second data being directed to a second storage volume, writing the first data, as part of a first I/O (input/output) process which begins before a selected time, to a first storage image and a second storage image, the first storage image and the second storage image forming a data mirror prior to the selected time, wherein writes to one image are replicated to the other image, and writing the second data, as part of a second I/O process which begins after the selected time, to the second storage image but not to the first storage image, the second I/O process being capable of running while the first process runs. Other methods and apparatuses are also described.
摘要:
A storage array interconnection fabric may be configured using a torus topology. A storage system including a path-redundant torus interconnection fabric is coupled to a plurality of nodes. The torus interconnection fabric may be configured to connect the plurality of nodes in an array including N rows and M columns, where N and M are positive integers. The array may be configured such that a first node in a first row of the N rows is connected to a second node in the first row and a first node in a first column of the M columns is connected to a second node in the first column. Also an ending node in the first row is connected to the first node in the first row and an ending node in the first column is connected to the first node in the first column. In addition, a first portion of the plurality of nodes is configured to communicate with a plurality of storage devices such as disk drives.
摘要:
One embodiment of a storage controller is described including a controller memory, one or more central processing units (CPUs), and a host bus adapter all coupled to a controller bus. The one or more CPUs are configured to produce data routing information dependent upon a data transfer command which directs a transfer of data between a host computer and one or more storage devices. The host bus adapter includes a receive unit and a transmit unit adapted for coupling to a transmission medium. The host bus adapter receives the data routing information, and forwards data associated with the data transfer command from the receive unit to the transmit unit dependent upon the data routing information such that the data associated with the data transfer command is not conveyed upon the controller bus and is not stored within the controller memory. As a result, the data transfer mechanism does not rely upon availability of the controller bus and/or the one or more CPUs, allowing independent scalability of input/output operations per second (IOPS) and data transfer rate of a storage system including the storage controller. Embodiments of storage and computer systems including the storage controller are also described.
摘要:
A variably positionable electrical connector provides a direct interface for a disk drive or other electrical device to a printed circuit board (PCB), backplane or motherboard of a computer system. The connector has a base (which may comprise a PCB or backplane) and a housing that slides relative to the base to allow the housing to be positioned according to the spacing between the electrical device's power and signal connectors. The housing includes multiple electrical contacts that receive or engage corresponding contacts of the device. Conductors that are electrically coupled to the computer system extend from the base and include portions that are aligned substantially parallel to a direction in which the housing can slide. The housing contacts slidably engage the parallel portions of the conductors and, as the housing is moved, the housing contacts slidably maintain electrical contact with the conductors. Either or both of the computer system's power and signal connectors that engage the device's connectors may be variably positionable, with the conductors of the variably positionable connector being coupled either to a power supply or a signal source (e.g., processor, bus, memory) as appropriate.
摘要:
A fault tolerant memory control system is provided for a computer system having a host processor, a memory and a system interconnect. The memory control system includes a primary memory controller and a backup memory controller with a tap coupled to the interconnect. Data is transferred from the host processor to the memory in the form of data packets. First, the host processor writes to the memory by sending a data packet to the primary memory controller which then caches the data from the data packet. The backup memory controller taps the interconnect to obtain a backup copy of the data packet as the data packet is being sent from the host processor to the primary memory controller which caches the data from the backup copy of the data packet. If the primary memory controller is functional, the primary memory controller sends the data to the memory via a primary path coupling the primary memory controller to the memory. Conversely, if the primary memory controller fails, i.e., is non-functional, the backup memory controller is tasked with completing the data transfer via a backup path coupling the backup memory controller to the memory.
摘要:
Broadly speaking, an apparatus for efficiently utilizing a shared packet buffer memory in a switch and a method for operating the same is provided. More specifically, the apparatus includes a memory having a number of buffers configured to be operated in a ratcheted manner. The ratcheted manner in which the memory is operated causes each incoming data stream to be distributed across the number of buffers. Each stored data stream can also be retrieved from the number of buffers for output from the memory in a similar ratcheted manner. The memory uses a rotating selector to control the ratcheted manner of operation. Also, the memory is capable of simultaneously servicing each of a number of inputs and a number of outputs to which the memory is connected.
摘要:
A single field replaceable storage or computer system may include a processor coupled to a peripheral bus by a bridge device. The field replaceable unit (FRU) may also include system memory coupled to the processor and a network interface coupled to the peripheral bus. One or more drive controllers may also be included coupled to the peripheral bus. Additionally, the single field replaceable unit includes an array of disk drives coupled to the one or more drive controllers. The array of disk drives may be configured as one or more RAID logical volumes and exported or presented to client machines as one or more file systems through the network interface. The processor, system memory, network interface, drive controllers, and array of disk drives are all packaged together as a single field replaceable unit. The processor, system memory, network interface, drive controllers, and array of disk drives may be configured not to be individually field serviceable or replaceable. If the unit fails and needs servicing or replacement, the unit may be replaced as a whole.
摘要:
Various methods and systems for cooling high-density arrangements of disk drives are disclosed. A disk drive enclosure includes several stacked layers of disk drives and one or more air movers. Some of the air movers are configured to cool the disk drives by creating an airflow. The disk drives are configured to operate as a network or computer storage system. Instead of being arranged in a traditional, aligned arrangement, the disk drives are arranged in an offset or staggered arrangement so that at least one disk drive in a first layer is offset from an overlapping disk drive in a second layer. The offset is in a direction parallel to the plane that includes the first layer. As a result, at least part of one of the disk drives in the arrangement is exposed to more of the airflow than it would be exposed to in an aligned arrangement.