Abstract: A method and an apparatus of allocating available resources in a cluster system with learning models and tuning methods are provided. The learning model may be trained from historic performance data of previously executed jobs and used to project a suggested amount of resources for execution of a job. The tuning process may suggest a configuration for the projected amount of resources in the cluster system for an optimal operating point. An optimization may be performed with respect to a set of objective functions to improve resource utilization and system performance while suggesting the configuration. Through many executions and job characterization, the learning/tuning process for suggesting the configuration for the projected amount of resources may be improved by understanding correlations of historic data and the objective functions.

Abstract: A method and device for provisioning bare-metal servers to tenants of a cloud provider is provided. The cloud provider operating a datacenter receives a profile of a tenant. The cloud provider identifies a set of physical resources subscribed by the tenant based on the tenant's profile and identifies a bare-metal server in the datacenter having sufficient physical resources to provide the subscribed physical resources. The cloud center reports to the tenant a set of physical resources in the BM server that corresponds to the subscribed set of physical resources. The cloud center receives a request from the tenant to provision a tenant bare metal machine in the datacenter. The cloud center enables at least one of the reported physical resources in the bare-metal server for the tenant, based on the request, and disables remaining reported physical resources in the bare-metal.

Abstract: The present disclosure relates to sharing an I/O device across multiple virtual machines. According to one embodiment, a computing system configures shared ownership of the I/O device between a first partition and one or more of the plurality of virtual machines. The computing system transfers partial ownership of the I/O device from the first partition to the one or more virtual machines and generates device configuration information for the I/O device. The virtual machines can use the generated device configuration information to access and configure the I/O device. Once the I/O device is configured for shared ownership, the computing system boots the one or more virtual machines.

Abstract: An apparatus for protecting connectors protruding from a metal enclosure of a hardware canister includes an enclosure that includes one or more components, wherein a portion of the one or more components protrudes out of a first side of the enclosure. The apparatus further includes a first extendable member assembly coupled to a lower interior surface of the enclosure, wherein the first extendable member assembly includes, a first structure coupled to a first compressible member, wherein a force applied to the first structure at the first side of the enclosure translates to the first compressible member and a lower edge of the first structure located below a lower edge of the one or more components, wherein a length of the first structure in an extended state is longer than a length of the protruding portion of the one or more components.

Abstract: An apparatus for protecting connectors protruding from a metal enclosure of a hardware canister includes an enclosure that includes one or more components, wherein a portion of the one or more components protrudes out of a first side of the enclosure. The apparatus further includes a first extendable member assembly coupled to a lower interior surface of the enclosure, wherein the first extendable member assembly includes, a first structure coupled to a first compressible member, wherein a force applied to the first structure at the first side of the enclosure translates to the first compressible member and a lower edge of the first structure located below a lower edge of the one or more components, wherein a length of the first structure in an extended state is longer than a length of the protruding portion of the one or more components.

Abstract: A method and device for provisioning bare-metal servers to tenants of a cloud provider is provided. The cloud provider operating a datacenter receives a profile of a tenant. The cloud provider identifies a set of physical resources subscribed by the tenant based on the tenant's profile and identifies a bare-metal server in the datacenter having sufficient physical resources to provide the subscribed physical resources. The cloud center reports to the tenant a set of physical resources in the BM server that corresponds to the subscribed set of physical resources. The cloud center receives a request from the tenant to provision a tenant bare metal machine in the datacenter. The cloud center enables at least one of the reported physical resources in the bare-metal server for the tenant, based on the request, and disables remaining reported physical resources in the bare-metal.

Abstract: Method and apparatus for performing in-place compression is provided. The in-place compression system transfers source data from a partition of a memory to a data buffer based on a read address. Compressed data is created by referencing the source data stored in the data buffer. The system writes the compressed data to the memory partition based on a write address. When the write address points at an address location that stores source data that has not been transferred to the data buffer, the system overwrites the compressed data stored in the memory partition with the source data stored in the data buffer.

Abstract: An apparatus for protecting connectors protruding from a metal enclosure of a hardware canister includes an enclosure that includes one or more components, wherein a portion of the one or more components protrudes out of a first side of the enclosure. The apparatus further includes a first extendable member assembly coupled to a lower interior surface of the enclosure, wherein the first extendable member assembly includes, a first structure coupled to a first compressible member, wherein a force applied to the first structure at the first side of the enclosure translates to the first compressible member and a lower edge of the first structure located below a lower edge of the one or more components, wherein a length of the first structure in an extended state is longer than a length of the protruding portion of the one or more components.

Abstract: Systems, methods, and computer program products to perform an operation comprising executing a device driver in a private logical partition on a compute host, wherein the device driver is configured to execute in an environment different than an environment of a hypervisor of the compute host, establishing a communication channel between the private logical partition and an adjunct partition executing on the compute host, and configuring, responsive to a command sent by the adjunct partition to the device driver via the communication channel, a physical function of a single root I/O virtualization (SR-IOV) device of the host system.

Abstract: A method and device for provisioning bare-metal servers to tenants of a cloud provider is provided. The cloud provider operating a datacenter receives a profile of a tenant. The cloud provider identifies a set of physical resources subscribed by the tenant based on the tenant's profile and identifies a bare-metal server in the datacenter having sufficient physical resources to provide the subscribed physical resources. The cloud center reports to the tenant a set of physical resources in the BM server that corresponds to the subscribed set of physical resources. The cloud center receives a request from the tenant to provision a tenant bare metal machine in the datacenter. The cloud center enables at least one of the reported physical resources in the bare-metal server for the tenant, based on the request, and disables remaining reported physical resources in the bare-metal.

Abstract: An apparatus for protecting connectors protruding from a metal enclosure of a hardware canister includes an enclosure that includes one or more components, wherein a portion of the one or more components protrudes out of a first side of the enclosure. The apparatus further includes a first extendable member assembly coupled to a lower interior surface of the enclosure, wherein the first extendable member assembly includes, a first structure coupled to a first compressible member, wherein a force applied to the first structure at the first side of the enclosure translates to the first compressible member and a lower edge of the first structure located below a lower edge of the one or more components, wherein a length of the first structure in an extended state is longer than a length of the protruding portion of the one or more components.

Abstract: Method and apparatus for performing in-place compression is provided. The in-place compression system transfers source data from a partition of a memory to a data buffer based on a read address. Compressed data is created by referencing the source data stored in the data buffer. The system writes the compressed data to the memory partition based on a write address. When the write address points at an address location that stores source data that has not been transferred to the data buffer, the system overwrites the compressed data stored in the memory partition with the source data stored in the data buffer.

Abstract: A method and device for provisioning bare-metal servers to tenants of a cloud provider is provided. The cloud provider operating a datacenter receives a profile of a tenant. The cloud provider identifies a set of physical resources subscribed by the tenant based on the tenant's profile and identifies a bare-metal server in the datacenter having sufficient physical resources to provide the subscribed physical resources. The cloud center reports to the tenant a set of physical resources in the BM server that corresponds to the subscribed set of physical resources. The cloud center receives a request from the tenant to provision a tenant bare metal machine in the datacenter. The cloud center enables at least one of the reported physical resources in the bare-metal server for the tenant, based on the request, and disables remaining reported physical resources in the bare-metal.

Abstract: A method and an apparatus of allocating available resources in a cluster system with learning models and tuning methods are provided. The learning model may be trained from historic performance data of previously executed jobs and used to project a suggested amount of resources for execution of a job. The tuning process may suggest a configuration for the projected amount of resources in the cluster system for an optimal operating point. An optimization may be performed with respect to a set of objective functions to improve resource utilization and system performance while suggesting the configuration. Through many executions and job characterization, the learning/tuning process for suggesting the configuration for the projected amount of resources may be improved by understanding correlations of historic data and the objective functions.

Abstract: Systems, methods, and computer program products to perform an operation comprising executing a device driver in a private logical partition on a compute host, wherein the device driver is configured to execute in an environment different than an environment of a hypervisor of the compute host, establishing a communication channel between the private logical partition and an adjunct partition executing on the compute host, and configuring, responsive to a command sent by the adjunct partition to the device driver via the communication channel, a physical function of a single root I/O virtualization (SR-IOV) device of the host system.

Abstract: The present disclosure relates to sharing an I/O device across multiple virtual machines. According to one embodiment, a computing system establishes a communication channel between a first partition and a second partition. The first partition generally owns an I/O device and the second partition hosts a device driver for the I/O device. The computing system configures shared ownership of the I/O device between the first partition and one or more virtual machines and transfers partial ownership of the I/O device to the second partition. Device configuration information is generated for the I/O device, which is used by the one or more virtual machines and the second partition to access and configure the I/O device. Subsequently, the computing system boots the one or more virtual machines.

Abstract: The present disclosure relates to sharing an I/O device across multiple virtual machines. According to one embodiment, a computing system configures shared ownership of the I/O device between a first partition and one or more of the plurality of virtual machines. The computing system transfers partial ownership of the I/O device from the first partition to the one or more virtual machines and generates device configuration information for the I/O device. The virtual machines can use the generated device configuration information to access and configure the I/O device. Once the I/O device is configured for shared ownership, the computing system boots the one or more virtual machines.

Abstract: A method, system, and computer program product for connection state recovery of a connection after fault in a networked channel-to-channel computer system are provided. The method includes identifying essential data in response to detecting a state change in a channel of the computer system, the essential data including connection state information used in performing a recovery operation. The method also includes separating the essential data from transient or incidental data, augmenting the essential data with validation data, and storing the augmented essential data in a memory location of a control unit in communication with the channel. In response to initiation of a recovery operation, the method includes retrieving the augmented essential data and validating contents of persistent data fields containing the essential data. The method further includes using the contents to restore the connection to an operational state when it is determined that the contents of the persistent data fields are valid.

Abstract: An apparatus and method for controlling power usage in a computer includes a plurality of computers communicating with a local control device, and a power source supplying power to the local control device and the computer. A plurality of sensors communicate with the computer for ascertaining power usage of the computer, and a system control device communicates with the computer for controlling power usage of the computer.

Abstract: A method, system, and computer program product for connection state recovery of a connection after fault in a networked channel-to-channel computer system are provided. The method includes identifying essential data in response to detecting a state change in a channel of the computer system, the essential data including connection state information used in performing a recovery operation. The method also includes separating the essential data from transient or incidental data, augmenting the essential data with validation data, and storing the augmented essential data in a memory location of a control unit in communication with the channel. In response to initiation of a recovery operation, the method includes retrieving the augmented essential data and validating contents of persistent data fields containing the essential data. The method further includes using the contents to restore the connection to an operational state when it is determined that the contents of the persistent data fields are valid.