Snowflake Inc.

Johan Harjono

Kunal Prafulla Nabar

Benoit Dageville

Steven James Pelley

Simon Holm Jensen

Daniel Geoffrey Karp

Arthur Kelvin Shi

Rares Radut

Thanakul Wattanawong

Samir Rehmtulla

Systems, methods, and devices for retrying a query. A method includes receiving a query directed to database data and assigning execution of the query to one or more execution nodes of an execution platform, the one or more execution nodes configured to execute the query on a first version of a database platform. The method includes determining that execution of the query was unsuccessful. The method includes assigning a first retry execution of the query to the one or more execution nodes of the execution platform and determining whether a regression or an intermittent fault caused the execution of the query to be unsuccessful based at least in part on whether the first retry execution of the query was successful or unsuccessful.

Identifying software regressions based on query retry attempts in a database environment

Techniques described herein can optimize usage of computing resources in a data system. Dynamic throttling can be performed locally on a computing resource in the foreground and autoscaling can be performed in a centralized fashion in the background. Dynamic throttling can lower the load without overshooting while minimizing oscillation and reducing the throttle quickly. Autoscaling may involve scaling in or out the number of computing resources in a cluster as well as scaling up or down the type of computing resources to handle different types of situations.

Autoscaling and throttling in an elastic cloud service

Techniques for automated query retry in a database platform include decoding, by at least one hardware processor, a query directed to database data. The query is received by a first management node from a client account of a database system. The method further includes configuring an execution of the query by a first computing node associated with the first management node. The method includes detecting the execution of the query results in a failed execution. The query is transferred from the first management node to a second management node based on the failed execution. The second management node is configured to process tasks that are internal to the database system. A retry execution of the query by a second computing node is configured. The second computing node is associated with the second management node. A cause of the failed execution is determined based on a result of the retry execution.

Automated query retry execution using multiple computing nodes in a database system

Automated query retry includes selecting a first node of a plurality of nodes of an execution platform in response to a query. The first node is configured with a first version of a database platform. A first execution of the query is performed using the first version of the database platform at the first node. The method further includes detecting the first execution results in a failed execution. A second execution of the query is scheduled. The second execution uses at least one of the first version of the database platform or a second version of the database platform configured at a second node of the plurality of nodes. The second version is released after the first version. A cause of the failed execution of the query at the first node is determined based at least in part on a result of the second execution of the query.

Automated query retry using a database platform

The subject technology selects a particular zone among multiple zones based on a target skew to meet a global balancing of cluster instances. The subject technology deploys a particular type of cluster instance to the particular zone. The subject technology, for each zone from the multiple zones, determines a respective number of cluster instances. The subject technology identifies a second particular type of cluster instance to add based on a total number of the second particular type of cluster instance in the multiple zones and a second total number of the particular type of cluster instance in the multiple zones. The subject technology adds the second particular type of cluster instance to a second particular zone to meet the global balancing of cluster instances in the multiple zones.

Cluster balancing for zones of a database system

The subject technology selects a particular zone among multiple zones based on a target skew to meet a global balancing of cluster instances. The subject technology deploys a particular type of cluster instance to the particular zone. The subject technology, for each zone from the multiple zones, determines a respective number of cluster instances. The subject technology identifies a second particular type of cluster instance to remove based on a total number of the second particular type of cluster instance in the multiple zones and a second total number of the particular type of cluster instance in the multiple zones. The subject technology removes the second particular type of cluster instance from a second particular zone to meet the global balancing of cluster instances in the multiple zones.

Cluster instance balancing of a database system across zones

Disclosed herein are embodiments of systems and methods for query-attempt processing in a database environment. In an embodiment, a database platform receives a query, and operates a first query manager of the database platform to manage a first attempt to execute the query. Based on determining that the attempt to execute the query was unsuccessful, the database platform identifies a second query manager of the database platform as implementing a same version of the database platform as the first query manager, and operates the second query manager to manage a first retry attempt to execute the query.

Query-attempt processing in a database environment

Techniques for automated query retry in a database platform include decoding, by at least one hardware processor, a query directed to database data from a client account of a database platform. The method further includes decoding, by the at least one hardware processor, an indication that execution of the query on at least one computing node of the database platform results in a failed execution. The method further includes configuring a processing loop with continuous retry executions of the query on the at least one computing node based on the indication. The method further includes exiting the processing loop based on detecting a retry execution of the continuous retry executions results in at least one successful execution of the query. The method includes logging each attempt to execute the query during the continuous retry executions in a query status log until the at least one successful execution of the query.

Automated query retry execution in a database system

Techniques for automated query retry in a database platform include assigning by at least one hardware processor a first execution of a query directed to database data to a first execution node of a plurality of execution nodes of an execution platform. The first execution node uses a first set of configurations during the first execution. The techniques further include determining that the first execution of the query by the first execution node results in a failed execution. The query is transferred to a second execution node of the plurality of execution nodes. A second execution of the query at the second execution node is caused. The second execution node uses a second set of configurations during the second execution. A cause of the failed execution at the first execution node is determined based on a result of the second execution of the query at the second execution node.

Automated query retry in a database system

Autoscaling in an elastic cloud service

The subject technology determines an availability zone skew among multiple zones. The subject technology, based on the availability zone skew, determines a target skew to meet a global balancing of cluster instances. The subject technology, based on the target skew, selects a particular zone among multiple zones. The subject technology deploys a particular type of cluster instance to the particular zone. The subject technology, for each zone from the multiple zones, determining a respective number of cluster instances. The subject technology identifies a first zone that includes a highest number of cluster instances based on the respective number of cluster instances from each zone. The subject technology identifies a second zone that includes a lowest number of cluster instances based on the respective number of cluster instances from each zone.

Growing community of inventors

San Francisco, CA, United States of America

Johan Harjono