Abstract: Methods of machine learning for system deployments without performance regressions are performed by systems and devices. A performance safeguard system is used to design pre-production experiments for determining the production readiness of learned models based on a pre-production budget by leveraging big data processing infrastructure and deploying a large set of learned or optimized models for its query optimizer. A pipeline for learning and training differentiates the impact of query plans with and without the learned or optimized models, selects plan differences that are likely to lead to most dramatic performance difference, runs a constrained set of pre-production experiments to empirically observe the runtime performance, and finally picks the models that are expected to lead to consistently improved performance for deployment. The performance safeguard system enables safe deployment not just for learned or optimized models but also for additional of other ML-for-Systems features.

Abstract: Methods of machine learning for system deployments without performance regressions are performed by systems and devices. A performance safeguard system is used to design pre-production experiments for determining the production readiness of learned models based on a pre-production budget by leveraging big data processing infrastructure and deploying a large set of learned or optimized models for its query optimizer. A pipeline for learning and training differentiates the impact of query plans with and without the learned or optimized models, selects plan differences that are likely to lead to most dramatic performance difference, runs a constrained set of pre-production experiments to empirically observe the runtime performance, and finally picks the models that are expected to lead to consistently improved performance for deployment. The performance safeguard system enables safe deployment not just for learned or optimized models but also for additional of other ML-for-Systems features.

Abstract: Methods for optimization in query plans are performed by computing systems via a query optimizer advisor. A query optimizer advisor (QO-Advisor) is configured to steer a query plan optimizer towards more efficient plan choices by providing rule hints to improve navigation of the search space for each query in formulation of its query plan. The QO-Advisor receives historical information of a distributed data processing system as an input, and then generates a set of rule hint pairs based on the historical information. The QO-Advisor provides the set of rule hint pairs to a query plan optimizer, which then optimizes a query plan of an incoming query through application of a rule hint pair in the set. This application is based at least on a characteristic of the incoming query matching a portion of the rule hint pair.

Abstract: Methods of machine learning for system deployments without performance regressions are performed by systems and devices. A performance safeguard system is used to design pre-production experiments for determining the production readiness of learned models based on a pre-production budget by leveraging big data processing infrastructure and deploying a large set of learned or optimized models for its query optimizer. A pipeline for learning and training differentiates the impact of query plans with and without the learned or optimized models, selects plan differences that are likely to lead to most dramatic performance difference, runs a constrained set of pre-production experiments to empirically observe the runtime performance, and finally picks the models that are expected to lead to consistently improved performance for deployment. The performance safeguard system enables safe deployment not just for learned or optimized models but also for additional of other ML-for-Systems features.

Abstract: Aspects extend to methods, systems, and computer program products for optimally pipelining result sets with fault tolerance in distributed query execution. Distributed computing jobs are optimized by dividing the distributed computing jobs into one or more bubbles for execution. Each bubble can be independently executed, potentially in parallel with other bubbles, when resources to handle the bubble are available. Intra-bubble communication can be streamed between vertices within a bubble. Inter-bubble communication can be stored to durable storage. Bubbles provide a failure boundary for a job graph and re-executing a bubble along with storage of intermediate results in durable storage can be used to recover from failures. When a vertex inside a bubble fails, computation can resume by rescheduling the execution of the failed bubble from the durable inputs for that bubble. Durable storage provides a light-weight failover to handle non-deterministic behavior.

Abstract: Aspects extend to methods, systems, and computer program products for optimally pipelining result sets with fault tolerance in distributed query execution. Distributed computing jobs are optimized by dividing the distributed computing jobs into one or more bubbles for execution. Each bubble can be independently executed, potentially in parallel with other bubbles, when resources to handle the bubble are available. Intra-bubble communication can be streamed between vertices within a bubble. Inter-bubble communication can be stored to durable storage. Bubbles provide a failure boundary for a job graph and re-executing a bubble along with storage of intermediate results in durable storage can be used to recover from failures. When a vertex inside a bubble fails, computation can resume by rescheduling the execution of the failed bubble from the durable inputs for that bubble. Durable storage provides a light-weight failover to handle non-deterministic behavior.

Abstract: Methods, systems, and computer program products are provided for partitioning online databases. Online database operations, such as, for example, SPLIT, MERGE, and DROP, are used to alter the arrangement of partitions in a federated database. A SPLIT operation splits rows at one partition across a plurality of other partitions. A MERGE operation merges rows at a plurality of partitions in to one partition. A DROP operation shifts responsibility for rows of data from one partition to another partition and then drops the rows from the one partition.

Abstract: Methods, systems, and computer program products are provided for partitioning online databases. Online database operations, such as, for example, SPLIT, MERGE, and DROP, are used to alter the arrangement of partitions in a federated database. A SPLIT operation splits rows at one partition across a plurality of other partitions. A MERGE operation merges rows at a plurality of partitions in to one partition. A DROP operation shifts responsibility for rows of data from one partition to another partition and then drops the rows from the one partition.

Abstract: The present invention extends to methods, systems, and computer program products for scoping the context used to access a database partition. Embodiments of the invention enable data isolation using partitions in multi-tenant databases, while relieving client applications from dealing with the partitions. For example, a computer system that includes a distributed database system comprising a plurality of database partitions in a federation receives a context to use when performing database access operations within the distributed database system. The context identifies specified relevant portion of the federation. The computer system also receives a database access operation that is associated with the context. The computer system modifies the semantics of the database access operation in accordance with the associated context, to direct application of the database access operation to the specified relevant portion of the federation.

Abstract: Methods, systems, and computer program products are provided for partitioning online databases. Online database operations, such as, for example, SPLIT, MERGE, and DROP, are used to alter the arrangement of partitions in a federated database. A SPLIT operation splits rows at one partition across a plurality of other partitions. A MERGE operation merges rows at a plurality of partitions in to one partition. A DROP operation shifts responsibility for rows of data from one partition to another partition and then drops the rows from the one partition.

Abstract: The present invention extends to methods, systems, and computer program products for partitioning online databases. Online database operations, such as, for example, SPLIT, MERGE, and DROP, are used to alter the arrangement of partitions in a federated database. A SPLIT operation splits rows at one partition across a plurality of other partitions. A MERGE operation merges rows at a plurality of partitions in to one partition. A DROP operation shifts responsibility for rows of data from one partition to another partition and then drops the rows from the one partition.

Abstract: The present invention extends to methods, systems, and computer program products for scoping the context used to access a database partition. Embodiments of the invention enable data isolation using partitions in multi-tenant databases, while relieving client applications from dealing with the partitions. For example, a computer system that includes a distributed database system comprising a plurality of database partitions in a federation receives a context to use when performing database access operations within the distributed database system. The context identifies specified relevant portion of the federation. The computer system also receives a database access operation that is associated with the context. The computer system modifies the semantics of the database access operation in accordance with the associated context, to direct application of the database access operation to the specified relevant portion of the federation.

Abstract: The present invention extends to methods, systems, and computer program products for partitioning online databases. Online database operations, such as, for example, SPLIT, MERGE, and DROP, are used to alter the arrangement of partitions in a federated database. A SPLIT operation splits rows at one partition across a plurality of other partitions. A MERGE operation merges rows at a plurality of partitions in to one partition. A DROP operation shifts responsibility for rows of data from one partition to another partition and then drops the rows from the one partition.

Abstract: Systems and methods that generate specialized plans for compiling SQL queries. A plan generator component scans the query representation for parameter sensitive predicates and evaluates each predicate individually based on the parameter values. Accordingly, queries can be identified not only based on their structures, but also based on their parameter conditions. The specialized plans are more efficient for particular values, wherein queries that employ such values are optimally executed.

Abstract: An incrementally-scalable file system and method. The system architecture enables file systems to be scaled by adding resources, such as additional filers and/or file servers, without requiring that the system be taken offline or being known to client applications. The system also provides for load balancing file accesses by distributing files across the various file storage resources in the system, as dictated by the relative capacities of said storage resources. The system provides one or more “virtual” file system volumes in a manner that makes it appear to client applications that all of the file system's storage space resides on the virtual volume(s), while in reality the files may be stored on many more physical volumes on the filers and/or file servers in the system.

Abstract: Systems and methods that generate specialized plans for compiling SQL queries. A plan generator component scans the query representation for parameter sensitive predicates and evaluates each predicate individually based on the parameter values. Accordingly, queries can be identified not only based on their structures, but also based on their parameter conditions. The specialized plans are more efficient for particular values, wherein queries that employ such values are optimally executed.

Abstract: A virtual file system and method. The system architecture enables a plurality of underlying file systems running on various file servers to be “virtualized” into one or more “virtual volumes” that appear as a local file system to clients that access the virtual volumes. The system also enables the storage spaces of the underlying file systems to be aggregated into a single virtual storage space, which can be dynamically scaled by adding or removing file servers without taking any of the file systems offline and in a manner transparent to the clients. This functionality is enabled through a software “virtualization” filter on the client that intercepts file system requests and a virtual file system driver on each file server. The system also provides for load balancing file accesses by distributing files across the various file servers in the system, through migration of data files between servers.

Abstract: A virtual file system and method. The system architecture enables a plurality of underlying file systems running on various file servers to be “virtualized” into one or more “virtual volumes” that appear as a local file system to clients that access the virtual volumes. The system also enables the storage spaces of the underlying file systems to be aggregated into a single virtual storage space, which can be dynamically scaled by adding or removing file servers without taking any of the file systems offline and in a manner transparent to the clients. This functionality is enabled through a software “virtualization” filter on the client that intercepts file system requests and a virtual file system driver on each file server. The system also provides for load balancing file accesses by distributing files across the various file servers in the system, through migration of data files between servers.

Abstract: An incrementally-scalable database system and method. The system architecture enables database servers to be scaled by adding resources, such as additional servers, without requiring that the system be taken offline. Such scaling includes both adding one or more computer servers to a given server cluster, which enables an increase in database read transaction throughput, and adding one or more server clusters to the system configuration, which provides for increased read and write transaction throughput. The system also provides for load balancing read transactions across each server cluster, and load balancing write transactions across a plurality of server clusters.

Abstract: This invention provides methods to locate and plan the retrieval of data from networked information sources in response to a user query. The methods utilize descriptions of the information sources, the information domain of the sources, and of the query. The methods of this invention integrate both legacy systems and full relational databases with an efficient, domain-independent, query-planning algorithm, reason about the capabilities of different information sources, handle partial goal satisfaction i.e., gather as much data as possible when all that the user requested cannot be gathered, are both sound and complete, and are efficient.