System and method for distributed back-off in a database-oriented environment

摘要

Embodiments may include a system configured to, for each of multiple database partitions, generate a respective value of commit latency dependent on the latency of one or more previous commit operations directed to that database partition. The system may also be configured to, for a given work item that includes a commit operation to be initiated on a given database partition of said multiple database partitions, determine whether the value of the commit latency for the given database partition exceeds a configurable commit latency threshold. The system may be configured to, in response to determining that the value of the commit latency for the given database partition exceeds the commit latency threshold, postpone the given work item to reduce load on the database partitions and perform the given work item after the postponement. If the commit latency does not exceed the threshold, the work may be performed without the postponement.

主权项

1. A system, comprising:
a database system comprising multiple database partitions configured to store data associated with one or more work items associated with processing requests made through a service interface; a plurality of host computers configured to perform work items requested through the service interface, wherein performing at least some of said work items includes initiating one or more commit operations on one or more of said database partitions, wherein at least some of the commit operations comprise updating a state of a transaction processed by the service; wherein one or more of the host computers are configured to generate, on a per-host, per-database partition basis, commit latency values for the database partitions, wherein a commit latency value for a particular database partition is dependent on the latency of one or more previous commit operations directed to that particular database partition; wherein a given one of the host computers is configured to:
determine, for a particular work item to be initiated on a respective one of the database partitions, that the given host computer's commit latency value for the respective database partition exceeds a commit latency threshold; andin response to said determine, reduce a load of work items on the respective database partition, wherein to reduce the load of work items, the given one of the host computers is configured to:
postpone the particular work item on the respective database partition; andperform the particular work item after the postponement.