Abstract: The present disclosure relates to systems, methods, and computer-readable media for peer booting server nodes on an edge zone. For example, an edge zone may include data plane devices capable of hosting virtual machines and other cloud-based services thereon that provide low-latency services to customers of a cloud computing system. Systems described herein implement an operating system (OS) provisioning system that, in combination with a peer booting system on a server node, facilitates instantiation of an OS on the server node as well as enabling the server node to act as an OS provider service to other server nodes on the edge zone. The systems disclosed herein facilitate fast installation of OS services without overutilizing limited bandwidth resources of a virtual network that connects the server nodes with control plane devices remotely maintained at a datacenter.

Abstract: The present disclosure describes a type of virtual machine, which the present disclosure may refer to as a harvest virtual machine, that may allow improved utilization of physical computing resources on a cloud-computing system. First, the harvest virtual machine may be evictable. In other words, higher priority virtual machines may preempt the harvest virtual machine's access to physical computing resources. Second, the harvest virtual machine may receive access to a dynamic amount of physical computing resources during the course of its operating life. Third, the harvest virtual machine may have a minimum size (in terms of an amount of physical computing resources) and may terminate whenever the harvest virtual machine has access to an amount of physical computing resources less than the minimum size.

Abstract: The present disclosure relates to systems, methods, and computer-readable media for pre-provisioning server nodes of a server rack and causing the pre-provisioned server node to be deployed on an edge zone of a cloud computing system. In particular, systems described herein involve identifying server nodes based on a customer hardware deployment request and performing a series of pre-provision acts to the server nodes in accordance with the received hardware deployment request. For example, systems described herein pre-provision the hardware by configuring hardware and software on the server nodes, establishing communication with one or more control planes on a datacenter, and bringing the server nodes to a return to web (RTW) state. By pre-provisioning the hardware, the server hardware may be delivered and transitioned to a live state in an efficient manner and without jeopardizing security on the cloud.

Abstract: Health and usage data are monitored for storage devices of a remote computing environment that is remote from a computing service provider. The remote computing environment comprises a subset of computing systems of a computing service provider that manages the remote computing environment. Based on the monitored health data, it is determined that one of the storage devices has failed. In response, it is determined whether a computing device of the remote computing environment that is associated with the failed storage device should be marked unhealthy or continue to operate at a reduced capacity. When it is determined that the computing device associated with the failed storage device should continue to operate at a reduced capacity, the computing device is reprovisioned with available storage devices and the computing device is operated at the reduced capacity.

Abstract: A cloud-computing system dynamically manages allocation of infrastructure nodes and compute nodes in an edge zone of the cloud-computing system. The edge zone begins with a first number of infrastructure nodes and a second number of compute nodes. As the edge zone executes customer workloads, the cloud-computing system determines whether the infrastructure nodes are over utilized or under utilized. When the infrastructure nodes are under utilized, the cloud-computing system re-assigns an infrastructure node to the compute nodes. When the infrastructure nodes are over utilized, the cloud-computing system re-assigns a compute node to the infrastructure nodes. In this way, the cloud-computing system dynamically maintains an optimal balance between resources devoted to supporting the edge zone (the infrastructure nodes) and resources devoted to executing customer workloads (the compute nodes).

Abstract: A system and method executed by a computer server include determination that the computer server has lost network connectivity and, in response to the determination that the computer server has lost network connectivity, disabling of power to a volatile memory of the computer server. Determining that the computer server has lost network connectivity may include transmission of data to an external system and determination that a response to the data was not received from the external system. Disabling of power to the volatile memory may include transmission of a request to an operating system of the system to initiate a power-down sequence of the system.

Abstract: Health and usage data are monitored for storage devices of a remote computing environment that is remote from a computing service provider. The remote computing environment comprises a subset of computing systems of a computing service provider that manages the remote computing environment. Based on the monitored health data, it is determined that one of the storage devices has failed. In response, it is determined whether a computing device of the remote computing environment that is associated with the failed storage device should be marked unhealthy or continue to operate at a reduced capacity. When it is determined that the computing device associated with the failed storage device should continue to operate at a reduced capacity, the computing device is reprovisioned with available storage devices and the computing device is operated at the reduced capacity.

Abstract: The present disclosure relates to systems, methods, and computer-readable media for peer booting server nodes on an edge zone. For example, an edge zone may include data plane devices capable of hosting virtual machines and other cloud-based services thereon that provide low-latency services to customers of a cloud computing system. Systems described herein implement an operating system (OS) provisioning system that, in combination with a peer booting system on a server node, facilitates instantiation of an OS on the server node as well as enabling the server node to act as an OS provider service to other server nodes on the edge zone. The systems disclosed herein facilitate fast installation of OS services without overutilizing limited bandwidth resources of a virtual network that connects the server nodes with control plane devices remotely maintained at a datacenter.

Abstract: Health and usage data are monitored for storage devices of a remote computing environment that is remote from a computing service provider. The remote computing environment comprises a subset of computing systems of a computing service provider that manages the remote computing environment. Based on the monitored health data, it is determined that one of the storage devices has failed. In response, it is determined whether a computing device of the remote computing environment that is associated with the failed storage device should be marked unhealthy or continue to operate at a reduced capacity. When it is determined that the computing device associated with the failed storage device should continue to operate at a reduced capacity, the computing device is reprovisioned with available storage devices and the computing device is operated at the reduced capacity.

Abstract: Health and usage data are monitored for storage devices of a remote computing environment that is remote from a computing service provider. The remote computing environment comprises a subset of computing systems of a computing service provider that manages the remote computing environment. Based on the monitored health data, it is determined that one of the storage devices has failed. In response, it is determined whether a computing device of the remote computing environment that is associated with the failed storage device should be marked unhealthy or continue to operate at a reduced capacity. When it is determined that the computing device associated with the failed storage device should continue to operate at a reduced capacity, the computing device is reprovisioned with available storage devices and the computing device is operated at the reduced capacity.

Abstract: A system and method executed by a computer server include determination that the computer server has lost network connectivity and, in response to the determination that the computer server has lost network connectivity, disabling of power to a volatile memory of the computer server. Determining that the computer server has lost network connectivity may include transmission of data to an external system and determination that a response to the data was not received from the external system. Disabling of power to the volatile memory may include transmission of a request to an operating system of the system to initiate a power-down sequence of the system.

Abstract: The present disclosure describes a type of virtual machine, which the present disclosure may refer to as a harvest virtual machine, that may allow improved utilization of physical computing resources on a cloud-computing system. First, the harvest virtual machine may be evictable. In other words, higher priority virtual machines may preempt the harvest virtual machine's access to physical computing resources. Second, the harvest virtual machine may receive access to a dynamic amount of physical computing resources during the course of its operating life. Third, the harvest virtual machine may have a minimum size (in terms of an amount of physical computing resources) and may terminate whenever the harvest virtual machine has access to an amount of physical computing resources less than the minimum size.

Abstract: Methods and system for collecting and disambiguating data across multiple, disparate publishers using a tracking uniform resource locator (URL) are provided. A tracking URL is created that maps a keyword or advertisement with a landing page. The tracking URL comprises a compound key and publisher specific formatting strings. A user selection corresponding to the tracking URL is received. The tracking URL along with fact data corresponding to the user selection is logged into a fact record. Further, the landing page associated with the tracking URL is retrieved and presented to the user.