Platform agnostic resource provisioning

• US 8,832,078 B2
• Filed: 10/02/2013
• Issued: 09/09/2014
• Est. Priority Date: 01/29/2013
• Status: Expired due to Fees

• Current Assignee: Tesora, Inc. (Stratoscale Ltd.)
• Original Assignee: Tesora, Inc. (Stratoscale Ltd.)
• Inventors: Annapragada, Mrithyunjaya, Jones, Morgan, Bramley, Andrew
• Primary Examiner(s): Saeed, Usmaan
• Assistant Examiner(s): Allen, Brittany N
• Application Number:
  US14/044,265
• Publication Number:
  US 20140214893A1
• Time in Patent Office: 342 Days
• Field of Search: 707/770, 707/719, 707/720
• US Class Current: 707/720
• CPC Class Codes: G06F 16/21 Design, administration or maintenance of databases
• CPC Class Codes: G06F 16/24542 Plan optimisation
• CPC Class Codes: G06F 16/24544 Join order optimisation
• CPC Class Codes: G06F 16/2471 Distributed queries
• CPC Class Codes: G06F 16/284 Relational databases
• CPC Class Codes: G06F 16/90335 Query processing
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques

Abstract

An elastic parallel database system where metadata is specified out-of-band during database operations via a set of augmentation rules. The rules are used to augment or modify commands received, and indicate whether they are to be used for specific connections on which they are received, for all client connections, or some other scope.

Claims

1. A database management system comprising

a) a network interface, for receiving database queries from two or more client applications, the client applications operating on two or more user client computers, the system providing a least one connection into the system for each such client application;

b) a group of one or more operational nodes for executing the queries as database operations, each operational node implemented as a logical collection of software components that execute on one or more physical machines, and one or more platforms;

c) client applications establish connections with the system and interact with the system in a query language;

d) one or more operational nodes operating as storage nodes and collectively referred to as a storage nodegroup;

e) one or more operational nodes operating as compute nodes and collectively referred to as compute nodegroup;

f) the system further storing user data in tables, and grouping the tables into databases;

g) tables are associated with one or more storage nodegroups and data for the table is stored on the nodes that are part of the associated storage nodegroup(s);

h) a Dynamic Query Planner (DQP) analyzes queries received for execution from applications and transforms those queries into query execution steps (QES) based on a query type, a schema of the referenced tables, and a manner in which data in the tables is distributed on the storage nodegroup(s);

i) a Query Execution Engine (QEE) executes these QES and produces results of a query that are then returned to the client that originated the query;

j) a persistent state is associated with each connection including at least a compute-nodegroup that is associated with the connection;

k) intermediate data streams generated as part of query processing related to a connection are associated with a nodegroup, including the compute-nodegroup(s) associated with the connection; and

l) one or more physical data processing machines providing compute elasticity by altering which compute nodes are part of the compute nodegroup associated with the connection, or by changing the compute nodegroup(s) associated with the connection, and provisioning and de-provisioning compute nodes dynamically, and further comprising;

a Compute Node Plugin which handles the provisioning and deprovisioning of instances, the provisioning and deprovisioning of compute nodes and their assignment to the QEE for executing specific queries, Policy Rules which determine the conditions under which, and the parameters within which the Compute Node Plugin may act, at least one platform on which the database executes, and at least two queries that share the same compute node during query execution.

2. The system of claim 1 wherein the QEE provisions and deprovisions compute nodes using Application Programming Interfaces (API'"'"'s) provided by the Compute Node Plugin.

3. The system of claim 1 wherein operations of the Compute Node Plugin are driven by the set of Policy Rules.

4. The system of claim 1 wherein the Compute Node Plugin interacts with the platform using API'"'"'s exposed by the platform.

5. The system of claim 1 wherein a QEE requests one or more Compute Nodes for the execution of a query through a Node Request API provided by the Compute Node Plugin, and upon receipt of such a request, the Compute Node Plugin consults a list of available nodes, and node utilization, and based upon the Policy Rules, provides the Query Execution Engine with compute nodes for it to use.

6. The system of claim 5 wherein the QEE, upon completion of a query, releases one or more Compute Nodes previously provisioned through a Node Request API by invoking a Node Release API, and upon receipt of a Node Release request, the Compute Node plugin records this action in a Node List.

7. The system of claim 1 wherein the Compute Node Plugin periodically reviews utilization of Compute Nodes provisioned from the platform and updates a Node Utilization list.

8. The system of claim 1 wherein the Compute Node Plugin provisions nodes from the platform(s) configured for execution of the query based on Policy Rules that consist of hard rules (absolutes) and soft rules (recommendations) and whereby the Compute Node Plugin is permitted to violate some of the soft rules but never permitted to violate any of the hard rules.

9. The system of claim 1 wherein the system provides a mechanism whereby the Policy Rules are changed dynamically either for the current connection, or for the system as a whole, or for some other scope specified by the query.

10. The system of claim 1 wherein the QEE makes API calls to the Compute Node Plugin irrespective of the underlying platform on which the Compute Nodes are to be provisioned and the same API calls are used to provision compute nodes irrespective of the underlying platform, providing however that parameters to the API calls specify either requirements or preferences for the platform to be chosen in provisioning compute nodes.

14. The method of claim 1 wherein the Compute Node Plugin is further for interacting with the platform using API'"'"'s exposed by the platform.

11. A method for operating a database management system comprising:

a) receiving database queries via a network interface from two or more client applications, the client applications operating on two or more user client computers, to provide a least one connection into the system for each such client application;

b) executing the queries as database operations on a group of one or more operational nodes, each operational node implemented as a logical collection of software components that execute on one or more physical machines, and one or more platforms;

c) the client applications interacting with the system in a query language;

d) operating one or more operational nodes as storage nodes and collectively referred to as a storage nodegroup;

e) operating one or more operational nodes as compute nodes and collectively referred to as a compute nodegroup;

f) storing user data in tables, and grouping the tables into databases;

g) associating tables with one or more storage nodegroups such that data for the tables is stored on the nodes that are part of the associated storage nodegroup(s);

h) analyzing queries via a Dynamic Query Planner (DQP) process and thereby transforming those queries into query execution steps (QES) based on a query type, a schema of the referenced tables, and a manner in which data in the tables is distributed on the storage nodegroup(s);

i) executing the QES via a Query Execution Engine (QEE) to produce results of a query that are then returned to the client that originated the query;

j) associating a persistent state with each connection including at least a compute-nodegroup that is associated with the connection;

k) generating intermediate data streams as part of query processing related to a connection and associating the intermediate data streams with a nodegroup, including the compute-nodegroup(s) associated with the connection; and

l) achieving compute elasticity by altering which compute nodes are part of the compute nodegroup associated with the connection, or by changing the compute nodegroup(s) associated with the connection, and provisioning and de-provisioning compute nodes dynamically from platform(s) on which the database system is executing, and further comprising;

handling, via a Compute Node Plugin, provisioning and deprovisioning of instances, the provisioning and deprovisioning of compute nodes and their assignment to the QEE for executing specific queries, Policy Rules which determine the conditions under which, and the parameters within which the Compute Node Plugin may act, at least one platform on which the database executes, and at least two queries that share the same compute node during query execution.

12. The method of claim 11 wherein the QEE is further for provisioning and deprovisioning compute nodes using Application Programming Interfaces (API'"'"'s) provided by the Compute Node Plugin.

13. The method of claim 11 additionally comprising:

driving operations of the Compute Node Plugin are driven by the set of Policy Rules.

15. The method of claim 11 wherein a QEE is further for requesting one or more Compute Nodes for the execution of a query through a Node Request API provided by the Compute Node Plugin, and upon receipt of such a request, the Compute Node Plugin is further for consulting a list of available nodes, and node utilization, and based upon the Policy Rules, providing the Query Execution Engine with compute nodes for it to use.

16. The method of claim 15 wherein the QEE, upon completion of a query, is further for releasing one or more Compute Nodes previously provisioned through a Node Request API by invoking a Node Release API, and upon receipt of a Node Release request, the Compute Node plugin is further for recording this action in a Node List.

17. The method of claim 11 wherein the Compute Node Plugin further is for periodically reviewing utilization of Compute Nodes provisioned from the platform and updating a Node Utilization list.

18. The method of claim 11 wherein the Compute Node Plugin is further for provisions nodes from the platform(s) configured for execution of the query based on Policy Rules that consist of hard rules (absolutes) and soft rules (recommendations) and whereby the Compute Node Plugin is further permitted to violate some of the soft rules but never permitted to violate any of the hard rules.

19. The method of claim 11 further comprising changing the Policy Rules changed dynamically either for the current connection, or for the system as a whole, or for some other connection scope specified by the query.

20. The method of claim 11 wherein the QEE is further for making API calls to the Compute Node Plugin irrespective of the underlying platform on which the Compute Nodes are to be provisioned and the same API calls are used to provision compute nodes irrespective of the underlying platform, providing however that parameters to the API calls specify either requirements or preferences for the platform to be chosen in provisioning compute nodes.