Amazon Technologies, Inc.

Atle Normann Jorgensen

Tate Andrew Certain

Matthew Scott Garman

Roland Alexander Paterson-Jones

Peter Nicholas Desantis

Carl Bellingan

Peter N DeSantis

Colin Laird Lazier

Marcin Piotr Kowalski

Matthew James Eddey

Anton André Eicher

André Mostert

Roland Paterson-Jones

Techniques are described for managing access of executing programs to non-local block data storage. In some situations, a block data storage service uses multiple server storage systems to reliably store copies of network-accessible block data storage volumes that may be used by programs executing on other physical computing systems, and snapshot copies of some volumes may also be stored (e.g., on remote archival storage systems). A group of multiple server block data storage systems that store block data volumes may in some situations be co-located at a data center, and programs that use volumes stored there may execute on other computing systems at that data center, while the archival storage systems may be located outside the data center. The snapshot copies of volumes may be used in various ways, including to allow users to obtain their own copies of other users' volumes (e.g., for a fee).

Providing executing programs with access to stored block data of others

Techniques are described for managing access of executing programs to non-local block data storage. In some situations, a block data storage service uses multiple server storage systems to reliably store network-accessible block data storage volumes that may be used by programs executing on other physical computing systems. A group of multiple server block data storage systems that store block data volumes may in some situations be co-located at a data center, and programs that use volumes stored there may execute on other physical computing systems at that data center. If a program using a volume becomes unavailable, another program (e.g., another copy of the same program) may in some situations obtain access to and continue to use the same volume, such as in an automatic manner in some such situations.

Managing access of multiple executing programs to non-local block data storage

Managing access of multiple executing programs to nonlocal block data storage

Embodiments are disclosed for determining whether a computing node is in a normal or an abnormal condition based on its characteristics relative to those of other computing nodes. In embodiments, characteristics about computing nodes (e.g., CPU resources used) are used to cluster those computing nodes, and those computing nodes that lie outside of a cluster are identified as abnormal.

Using unsupervised learning to monitor changes in fleet behavior

Systems and methods for performing diagnostic tests within a multi-tenant environment are described. Diagnostics are performed on one or more components, such as host computing devices. The one or more components send resulting diagnostic information to an ingester which collects the diagnostic information. The ingester then sends the diagnostic information to a database. A generator is able to query the information in the database and create metrics. Based at least in part on metrics, an alarm may be triggered.

Testing framework for host computing devices

Providing access to stored computing snapshots

Embodiments are disclosed for determining whether a computing node is in a normal or an abnormal condition based on its characteristics relative to those of other computing nodes. In embodiments, log files for the computing node are used to develop a state model of the computing node, and where the state model differs between two similar computing nodes, an abnormality is identified. In other embodiments, characteristics about computing nodes (e.g., CPU resources used) are used to cluster those computing nodes, and those computing nodes that lie outside of a cluster are identified as abnormal.

Determining abnormal conditions of host state from log files through Markov modeling

Techniques are described for enabling a virtual machine to be presented with an amount of available guest memory, where a hypervisor or other privileged component manages the mapping of the guest memory to either volatile memory (e.g., RAM) or to secondary storage (e.g., SSD). This enables volatile memory to be effectively oversubscribed to on host computing devices that have a limited amount of total available volatile memory but which are running multiple virtual machines. For example, each virtual machine on the device can be presented as having access to the total amount of available RAM that is available on the device. The hypervisor or other virtualization component then monitors the usage of the memory by each virtual machine and shapes which portions of the guest memory for that virtual machine are mapped to RAM and which portions are mapped to secondary storage, such as SSD.

Managing memory in virtualized environments

In this disclosure, a resource scheduler is described that allows virtual machine instances to earn resource credits during the low activity levels. Virtual machine instances that spend a predominant amount of time operating at low activity levels are able to quickly gain resource credits. Once these virtual machine instances acquire enough resource credits to surpass a threshold level, the resource scheduler can assign a high priority level to the virtual machine instances that provide them with priority access to CPU resources. The next time that the virtual machine instances enter a high activity level, they have a high priority level that allows them to preempt other, lower priority virtual machine instances. Thus, these virtual machine instances are able to process operations and/or respond to user requests with low latency.

Growing community of inventors

Cape Town, South Africa

Atle Normann Jorgensen