摘要:
A system that includes multiple hosts, each having physical resources, a hypervisor, and a host agent that causes the virtual machines to operate above the hypervisor so as to be abstracted from the physical resources. A host fabric reviews requests to instantiate or upgrade a virtual machine, identifies a corresponding virtual machine image to instantiate the virtual machine from in order to honor the request, and determines whether the virtual machine to be instantiated has a valid lease on the virtual machine image. If the requestor does not have a valid lease, the request is blocked. If the requestor does have a valid lease, the host fabric facilitates the instantiation of the virtual machine from the virtual machine image. This prevents or eliminates the chances of multiple instances of the same virtual machine image writing to the virtual machine image at the same time.
摘要:
Goal state indicators can be communicated from a fabric controller of a computer cluster to each of multiple compute instances in the computer cluster managed by the fabric controller. The goal state indicators can be formatted according to a structured protocol that defines multiple possible goal states. Additionally, status reports can be received from the compute instances. Each of the status reports can indicate a current state of one of the compute instances relative to a goal state previously indicated in a goal state indicator communicated to that one of the compute instances.
摘要:
Methods, systems, and computer-readable media for facilitating coordination between a fabric controller of a cloud-computing network and a service application running in the cloud-computing network are provided. Initially, an update domain (UD) that includes role instance(s) of the service application is selected, where the service application represents a stateful application is targeted for receiving a tenant job executed thereon. The process of coordination involves preparing the UD for execution of the tenant job, disabling the role instance(s) of the UD to an offline condition, allowing the tenant job to execute, and restoring the role instance(s) to an online condition upon completing execution of the tenant job. Preparing the UD includes notifying a management role established within the service application of the fabric controller's intent to execute the tenant job and receiving a management-role response communicating a presence of replications of internal state(s) of the role instance(s) affected by the tenant job.
摘要:
A form of multicast, referred to as “on-demand multicast,” provides a methodology that allows for “on demand” multicasting of data. On-demand multicast allows clients to join or leave a transmission at any point. There is no transmission session start or end time—the session is continuously available, although data is not necessarily continuously transmitted. Data blocks missed by late-joining clients are retransmitted per amalgamated client requests through multiple transmission cycles until all clients that have joined the transmission session receive all content. Once a client has received all data blocks comprising the content for the session, the client may exit the ongoing session.
摘要:
A control protocol is used to deploy and install an operating system image on a client. The data structure of the control protocol includes an operation code corresponding to an operation associated with the deployment and installation of the operating system image on the client. When a server receives a request packet from the client, the server parses the request packet and sends the parsed request packet to a provider. The provider then executes the operation associated with the operation code and generates a return value. The return value is sent to the server. The server then composes a reply packet including the return value and transmits the reply packet to the client.
摘要:
A computing device is configured to assemble and boot an operating system in stages. A primary boot image is first obtained that includes enough of the operating system to support basic networking. Once the primary boot image is obtained, either through download or from storage on the device, the primary boot image is booted to start the networking. Once the networking is active, the boot activity for the device is suspended and one or more secondary operating system images are downloaded by the device. These images are chained to the primary boot image thereby creating a single/cohesive operating system. These secondary operating system images are successively booted in “stages” with each successive stage building upon the functionality of the previous stage.
摘要:
Approving a client to boot in a pre-boot execution environment. A request is received from the client to boot the client by a computing device. The request includes device information associated with the client. A data store having a collection of known device information is queried in response to the received request. The request is stored, and it is determined whether to execute the received request by evaluating the device information in the stored request. The received request is responded to as a function of the evaluated device information.
摘要:
A method includes loading an optical media image representing the content of a physical optical media device from an optical media image source into random access memory (RAM) and emulating the physical optical media device using the optical media image. The optical media image is in an optical media format. A system includes an optical media image stored in random access memory (RAM), the optical media image being in an optical media format, and including operating system (OS) code executable by a microprocessor. A RAM disk program accesses the optical media image according to the optical media format.
摘要:
A form of multicast, referred to as “on-demand multicast,” provides a methodology that allows for “on demand” multicasting of data. On-demand multicast allows clients to join or leave a transmission at any point. There is no transmission session start or end time—the session is continuously available, although data is not necessarily continuously transmitted. Data blocks missed by late-joining clients are retransmitted per amalgamated client requests through multiple transmission cycles until all clients that have joined the transmission session receive all content. Once a client has received all data blocks comprising the content for the session, the client may exit the ongoing session.
摘要:
A slow client that participates in a reliable multicast session may reduce the speed of data transfer for other clients in the session. A multicast server can mitigate the effect of a slow client by identifying and disconnecting one or more clients that are slowing down the session, or by using other criteria to identify clients to be disconnected. The server may monitor the performance of clients in the session based on factors such as the rate at which each of the clients is able to receive data, the number or rate of packets lost by a client, the amount of time that a client has spent as the slowest client in the session, the identity of a particular client, etc. A client that has been identified for disconnection based on the server's detection efforts, or that has been identified by an administrator, may be disconnected.