International Business Machines Corporation

Maxim Kalaev

Zah Barzik

Rivka Mayraz Matosevich

Lior Chen

Amit Margalit

Jonathan Amit

Alexander Snast

Deborah A Messing

Vladislav Drouker

Eran Tzabari

Eli Koren

Dan Ben-Yaacov

Mor Griv

Yair Elharrar

Demian Glait

A mega cluster storage system includes clusters of multiple storage modules. Each module is able to access a portion of the data within the mega cluster and serves as a proxy in order for another storage module to access the remaining portion of the data. A cluster is assigned to a unique cluster volume and all the data within the cluster volume is accessible by all of the modules within the cluster. Each host connection to the mega cluster is associated with a particular cluster volume. A module that receives a host I/O request determines whether the I/O request should be satisfied by a module within its own cluster or be satisfied by a module within a different cluster. The module may forward the I/O request to a module within a different cluster as indicated by a distribution data structure that is allocated and stored within each storage module.

Layered clustered scale-out storage system

Improving storage device efficiency during data replication. An asynchronous data replication process is initiated between a source computer and a target computer over a network. Performance of the target computer and a storage device of the target computer is monitored. At least one portion of data stored on the source computer is identified for the data replication, wherein the at least one portion of the data is organized in a data block. Whether to modify a size of the data block is determined, based on the monitored performance of the target computer. Responsive to determining to modify the size of the data block, based on the monitored performance of the target computer, the size of the data block is modified.

Storage device efficiency during data replication

Embodiments for discovering read failures in a data storage system. Data read from underlying media of the data storage system is validated by initializing a prefix of a buffer used to read the data with a first sequence of bytes of a predefined number. The first sequence of bytes is compared to a sequence of actual bytes of the data read into the buffer. When the first sequence of bytes returns an equal value to the sequence of the actual bytes of the data read, a second sequence of bytes of the predefined number is generated and compared to a subsequent data read to determine the validity.

Efficient discovery of silent read failures on underlying media of storage systems

Methods, computing systems and computer program products implement embodiments of the present invention that include initializing multiple computing nodes to execute one or more software applications that provide a computing service, each of the computing nodes having a respective set of configuration data. Examples of the services include storage services, computational services and web service. Subsequent to deploying the multiple computing nodes as a software defined computing (SDC) cluster, an additional computing node is initialized using a given set of configuration data, and the additional computing node is added to the SDC cluster. In some embodiments, the additional node can be initialized and added to the SDC cluster in response to detecting a condition in the SDC cluster. In alternative embodiments, the additional node can be initialized and added to the SDC cluster in response to receiving a request to add an additional service to the SDC cluster.

Self-expanding software defined computing cluster

In one embodiment of the present invention, a server is created, and a computer readable storage media is included in the server. An adapter is connected to the server, a set of packets is received by the adapter from a network, and the network and a repository are interfaced by the adapter. A firmware is stored on the adapter, and the adapter is controlled by the firmware. A management component is stored on the computer readable storage media, and the server is managed by the management component. A driver is stored on the management component, and the driver communicates with the adapter. A set of filters that controls transmission of the set of packets is created, and each filter in the set of filters has a set of filter rules. The set of filters is stored on the server.

Network security for data storage systems

Embodiments for network management in a Software-defined Storage (SDS) environment. Remote mirroring connectivity and data transfer is optimized in the SDS environment by establishing each of a plurality of nodes in a local storage array and a remote storage array as a gateway node. Data residing on a local gateway node is transferred directly to a remote gateway node owning the data, thereby avoiding interconnect data hops and improving datacenter network utilization.

Efficient data transfer in remote mirroring connectivity on software-defined storage systems

Methods, systems, and computer program product embodiments for controlling congestion in a storage grid, by a processor device, are provided. In a storage grid, a storage request transmit queue length of storage requests submitted by a plurality of nodes is monitored, the storage request transmit queue comprising a queue holding the storage requests unable to be transmitted through a current transmission control protocol/internet protocol (TCP/IP) socket buffer of the storage grid. Upon reaching at least one of a certain threshold, a watermarked message indicating a specific one of the plurality of nodes is congested is transmitted to upper layer grid nodes in a hierarchy of the plurality of nodes transmitting the storage requests, and the upper layer grid nodes each alter the storage requests based upon the watermarked message.

Interconnect congestion control in a storage grid

Methods, computing systems and computer program products implement embodiments of the present invention that include defining, for a computer system, one or more system tests, and performing each of the system tests on the computer system. Each of the system tests includes one or more system commands, and in some embodiments, a given system test may be performed by executing one or more system commands while the computer system simultaneously processes additional system commands that perform input/output operations. Upon detecting a failure of a given system command in a given system test to execute correctly, the computer system can be configured to suppress execution of the given system command. In some embodiments, the computer system includes a hypervisor controlling a virtual machine configured as a node of a software defined storage system, and the failure can be a disruption of the hypervisor or the virtual machine.

Dangerous system command detection

Methods, systems, and computer program product embodiments for controlling congestion in a storage grid, by a processor device, are provided. In a storage grid, a storage request transmit queue length is monitored. Upon reaching at least one of a certain threshold, a watermarked message is transmitted to a receiving node, the receiving node altering storage requests based upon the watermarked message.

Various embodiments for method for detecting network and hardware failures in a computing environment, by a processor device, are provided. In one embodiment, a method comprises isolating component failures by deploying a diagnostic code within an Operating System (OS) of both a first node and a connected appliance, and executing a set of diagnostic tests using the diagnostic code to determine if the source of the failure is within the first node or the connected appliance.

Isolating hardware and network failures in a computing environment

Methods, computing systems and computer program products implement embodiments of the present invention that include initializing, by a processor executing a file system in communication with a block manager managing multiple storage regions on a storage device, a file system write cache to have a default cache size, the default cache size corresponding to a first storage capacity of a default number of the storage regions. Upon detecting that a current number of the storage regions that are not in use by the block manager is less than the default number, the file system write cache is resized to a reduced cache size that corresponds to a second storage capacity of the current number of the storage regions. While the file system write cache has the reduced cache size, the file system write cache can be resized back to the default cache size as unused storage regions become available.

Growing community of inventors

Petach Tikva, Israel

Maxim Kalaev