Seamless upgrades in a distributed database system

A computer program product for implementing a method for monitoring and maintaining a distributed database service upgrade process, the computer program product comprising one or more computer-readable storage media having stored thereon computer-executable instructions that, when executed by one or more processors of the computing system, cause the computing system to perform the method, the method comprising:

• an act of identifying one or more operating portions of a distributed database service that are to be monitored, wherein the identified operating portions represent potential points of failure for the distributed database service;
• an act of monitoring the identified operating portions to determine at least one of when and where a failure has occurred;
• an act of determining that one or more failures have occurred in one or more of the operating portions;
• and
• an act of evaluating the identified failures to determine if the computing system is capable of responding to the failures without outside intervention, wherein
• upon determining the computing system is capable of responding to the failures without outside intervention, the computing system automatically implements a determined response to resolve the failures and to allow the distributed database service to continue operating, or alternatively,
• upon determining the computing system is incapable of responding to the failures without outside intervention, the computing system temporarily pauses the distributed database service.

A distributed computer system comprising the following:

• one or more processors;
• system memory;
• one or more computer-readable storage media having stored thereon computer-executable instructions that, when executed by the one or more processors, causes the computing system to perform a method for providing distributed database service upgrades using multiple database instances, the method comprising the following:
• an act of instantiating, for each partition in the distributed database system, first and second server instances on a first node in a shared-nothing distributed computing system, first and second server instances on a first database secondary node in the shared-nothing distributed computing system, and first and second server instances of a second database secondary node in the shared-nothing distributed computing system, wherein the first instances are configured to operate using a current distributed database service version and the second instances are configured to operate using a new distributed database service version;
• an act of receiving an indication that the distributed database service is to be upgraded;
• and
• based on the received indication, an act of migrating each replica of each partition in the database, wherein the migration includes migrating from the first server instance of the primary node to the second server instance of the primary nodes on the first node in the shared-nothing distributed computing system, migrating the first server instance of the first secondary nodes to the second server instance of the first secondary node on the second nodes in the shared-nothing distributed computing system, and migrating the first server instance of the second secondary nodes to the second server instance of the second secondary nodes on the third node in the shared-nothing distributed computing system, substantially without user-visible downtime.