Amazon Technologies, Inc.

Adel Gawdat

Akhilesh Mritunjai

Richard David Krog

James Christopher Sorenson

Akshat Vig

Kai Zhao

Mostafa Elhemali

Wan Man Ryan Chan

Yogesh Kansal

Xi Huang

Tushar Rakheja

Jason Kroll

Richard Krog

Jonathan L Meed

Saumil Ramesh Hukerikar

Dolev Ish-am

Different types of index structures are used for a replica group of a database. A leader node of a replica group performs receives updates to a copy of the database using a first type of index structure. A follower node performs updates received from the leader node as a log of updates to a copy of the database in an external storage system when a size of the received updates exceeds a threshold. The follower node performs requests to read data from the database using the copy in the external storage.

Different types of index structures for storing database data in a replica group

Throughput capacity may be changed to sustain throughput for accessing individual items in a database. A table hosted at storage nodes that provide access to the table in a database may be identified as allocated with a client-specified throughput capacity for accessing the table. Performance of access requests to the table at the storage nodes may be tracked. Based on the performance of the access requests, a change may be determined that modifies a throughput capacity for the table to sustain a guaranteed throughput for each access request independent of other access requests received for the table.

Changing throughput capacity to sustain throughput for accessing individual items in a database

Techniques are described for monitoring and enabling access to information about key access patterns for nonrelational database tables managed by a nonrelational database service of a service provider network. Storage nodes store data as key-value pairs (or key-document pairs), where the records or documents of a database are addressed by a unique key for each record. The storage nodes over time receive requests to perform various operations (for example, to get, put, update, or delete data in a table) associated with records stored by the storage nodes. The storage nodes collect information indicating a frequency with which such requests are received for each key for which a storage node is responsible. This frequency information for each key can be normalized relative to a maximum frequency supported by the storage nodes to obtain an access frequency score, where the maximum frequency indicates a frequency at which additional requests to the storage node may be throttled.

Monitoring key access patterns for nonrelational databases

Growing community of inventors

Seattle, WA, United States of America

Adel Gawdat