Abstract: Disclosed are aspects of network function placement in virtual graphics processing unit (vGPU)-enabled environments. In one example a network function request is associated with a network function. A scheduler selects a vGPU-enabled GPU to handle the network function request. The vGPU-enabled GPU is selected in consideration of a network function memory requirement or a network function IO requirement. The network function request is processed using an instance of the network function within a virtual machine that is executed using the selected vGPU-enabled GPU.

Abstract: The disclosure provides an approach for creating a pool of memory out of local memories of host machines, and providing that pool for the hosts to use. The pool is managed by a controller that keeps track of memory usage and allocated memory among hosts. The controller allocates or reclaims memory between hosts, as needed by the hosts. Memory allocated from a second host to a first host may then be divided into smaller portions by the first host, and further allocated to virtual machines executing within the first host.

Abstract: Methods and apparatus to manage resources in a hybrid workload domain are disclosed. An example apparatus includes a usage monitor to monitor resource utilization of a workload allocated within a hybrid workload domain, and an orchestrator to: determine a first type of the workload domain in the hybrid workload domain; in response to determining that under-utilized resources of the first type are not available, identify resources of a second type that are available; convert the resources from the first type to the second type; and allocate the converted resources to the workload.

Abstract: In some embodiments, a method receives a set of packets for a flow and determines a set of features for the flow from the set of packets. A classification of an elephant flow or a mice flow is selected based on the set of features. The classification is selected before assigning the flow to a network resource in a plurality of network resources. The method assigns the flow to a network resource in the plurality of network resources based on the classification for the flow and a set of classifications for flows currently assigned to the plurality of network resources. Then, the method sends the set of packets for the flow using the assigned network resource.

Abstract: The disclosure provides an approach for detecting and preventing attacks in a network. Embodiments include receiving network traffic statistics of a system. Embodiments include determining a set of features of the system based on the network traffic statistics. Embodiments include inputting the set of features to a classification model that has been trained using historical features associated with labels indicating whether the historical features correspond to attacks. Embodiments include receiving, as output from the classification model, an indication of whether the system is a target of an attack. Embodiments include receiving additional statistics related to the system. Embodiments include analyzing, in response to the indication that the system is the target of the attack, the additional statistics to identify a source of the attack. Embodiments include performing an action to prevent the attack based on the source of the attack.

Abstract: Disclosed is a method, apparatus, and system for dynamically changing a backup policy, the operations comprising: automatically detecting a change in a backup source system; automatically activating a new backup policy, wherein the new backup policy is determined based on the change in the backup source system and an old backup policy; and performing a backup session based on the new backup policy.

Abstract: Various examples are disclosed for dynamic kernel slicing for virtual graphics processing unit (vGPU) sharing in serverless computing systems. A computing device is configured to provide a serverless computing service, receive a request for execution of program code in the serverless computing service in which a plurality of virtual graphics processing units (vGPUs) are used in the execution of the program code, determine a slice size to partition a compute kernel of the program code into a plurality of sub-kernels for concurrent execution by the vGPUs, the slice size being determined for individual ones of the sub-kernels based on an optimization function that considers a load on a GPU, determine an execution schedule for executing the individual ones of the sub-kernels on the vGPUs in accordance with a scheduling policy, and execute the sub-kernels on the vGPUs as partitioned in accordance with the execution schedule.

Abstract: A computing node of a data storage system includes a data transform component. The data transform component identifies data that is maintained by the storage system and has a particular data structure. The data transform component may also perform analytics on the data based on the data structure. For example, different data sets that are dependent, e.g. related via association with the same application or type of application, or which are typically accessed in temporal proximity, can be discovered based on data structure and analytics. Data relationships may be used to prepare dependent data for access in advance of an IO associated with that dependent data.

Abstract: Various examples are disclosed for dynamic kernel slicing for virtual graphics processing unit (vGPU) sharing in serverless computing systems. A computing device is configured to provide a serverless computing service, receive a request for execution of program code in the serverless computing service in which a plurality of virtual graphics processing units (vGPUs) are used in the execution of the program code, determine a slice size to partition a compute kernel of the program code into a plurality of sub-kernels for concurrent execution by the vGPUs, the slice size being determined for individual ones of the sub-kernels based on an optimization function that considers a load on a GPU, determine an execution schedule for executing the individual ones of the sub-kernels on the vGPUs in accordance with a scheduling policy, and execute the sub-kernels on the vGPUs as partitioned in accordance with the execution schedule.

Abstract: The disclosure provides an approach for creating a pool of memory out of local memories of host machines, and providing that pool for the hosts to use. The pool is managed by a controller that keeps track of memory usage and allocated memory among hosts. The controller allocates or reclaims memory between hosts, as needed by the hosts. Memory allocated from a second host to a first host may then be divided into smaller portions by the first host, and further allocated to virtual machines executing within the first host.

Abstract: Embodiments are described for detecting a data organization issue in a storage system by analyzing static and dynamic layout characteristics for data stored in the storage system, workload characteristics for applications utilizing the data, virtualization effects of the storage system on the organization of the data, input/output characteristics of processes storing the data in the storage system, and cache usage characteristics of the data. For fragmentation issues, the system uses the analysis to provide defragmentation advisories to reduce fragmentation by at least one of: re-layout or re-tiering of the data in the data storage system, or cache memory reconfiguration.

Abstract: Disclosed is a method, apparatus, and system for dynamically changing a backup policy, the operations comprising: automatically detecting a change in a backup source system; automatically activating a new backup policy, wherein the new backup policy is determined based on the change in the backup source system and an old backup policy; and performing a backup session based on the new backup policy.

Abstract: Techniques are disclosed for managing workloads across a distributed virtual infrastructure. In one example, a method comprises the following steps. Information is collected relating to services offered by service providers across a multi cloud environment implemented in accordance with a distributed virtual infrastructure. A service capability model is maintained based on the collected information. Service level claim conformance is validated for the services offered by the service providers. One or more recommendations are generated based on the collecting, maintaining and validating steps for presentation to a subscriber to assist in management of one or more workloads across the multi cloud environment.

Abstract: A system includes a metadata monitor configured to discover and manage metadata associated with a plurality of data protection mechanisms distributed and operating throughout a computing platform, and metadata representing a schema of the computing platform and one or more user applications executed thereon. The system also includes a data protection analytics manager comprising a set of one or more data protection analytics applications configured to determine, from the metadata, which resources are and are not protected by the plurality of data protection mechanisms. For the protected resources, the data protection analytics applications are configured to compute an effectiveness value, and for the non-protected resources, the data protection analytics applications are configure to compute an exposure value. A data protection model can then be adapted in a data protection mechanism-agnostic manner.

Abstract: Large files in a file system are mapped by extents of contiguous indirect blocks in order to reduce the time for read or write access to the large files while allowing allocation of data blocks one at a time. The inode of a file includes an indication of whether file mapping metadata of the file includes either an extent of contiguous indirect blocks or a tree of blocks. In a preferred mapping scheme, an inode contains an array of block pointer entries, and each entry includes a flag indicating whether the block pointer is pointing to either an extent of indirect blocks or a tree of blocks. For sharing of data blocks between files, the block pointer fields of the indirect blocks each contain an ownership flag indicating whether the pointed-to block is shared or not.

Abstract: A method, computer program product, and computing system for grouping storage blocks within a file system into a plurality of storage pools including a free-backed storage pool, a free-unbacked storage pool, and an allocated-backed storage pool. The free-backed storage pool identifies unused storage blocks within the file system that are already associated with physical storage space within a backend storage system. The free-unbacked storage pool identifies unused storage blocks within the file system that are not yet associated with physical storage space within the backend storage system. The allocated-backed storage pool identifies used storage blocks within the file system that are already associated with physical storage space within the backend storage system. A request is received for one or more unused storage blocks within the file system.

Abstract: Described are techniques for creating a snapshot of a distributed file system or a file thereof. The distributed file system includes metadata file systems storing metadata about files in the distributed file system and includes data file systems storing file content data. Quiescing is performed for requests affecting the distributed file system, or a file thereof, for which a snapshot is being created. A snapshot of metadata is obtained. For the distributed file system, this includes a snapshot of each metadata file system. For a file, this includes a snapshot of the file's metadata. A snapshot of file data is obtained. For distributed file systems, this includes a snapshot of each data file system. For a file, this includes a snapshot of the file's data. Unquiescing is performed for requests affecting the distributed file system, or the file thereof, upon successful completion of obtaining snapshots of metadata and file data.

Abstract: A storage processor having access to published logical unit numbers of a common block file system mounted on another storage processor processes requests for a particular file system object of the common block file system using metadata objects of the common block file system cached in both storage processors to access storage units shared between the two storage processors.

Abstract: A system and method for providing a faster disk recovery is provided by bypassing the file system cache temporarily holding a sub set of metadata objects of the file system and instead using a persistent fast storage that can be accessed at deterministic speeds to hold all the metadata objects of the file system. The system speeds recovery by only writing updated metadata objects to the persistent disk storage when file system recovery is complete.

Abstract: A dataset is recovered after a server reboot while clients access the dataset. In response to the reboot, not-yet-completed transactions in a log are parsed to create, for each of the dataset blocks modified by these active transactions, a respective block replay list of the active transactions that modify the block. Once the block replay lists have been created, clients may access specified blocks of the dataset after on-demand recovery of the specified blocks. The on-demand recovery is concurrent with a background recovery task that replays the replay lists. To accelerate log space recovery, the parsing of the log inserts each replay list into a first-in first-out queue serviced by multiple replay threads. The queue can also be used as the cache writeback queue, so that the cache index is used for lookup of the replay list and the recovery state of a given block.