Parallel processing database system with a shared metadata store
A method and system for executing database queries in parallel using a shared metadata store. The metadata store may reside on a master node, and the master node may distribute this metadata with query plans to worker nodes. The worker nodes may request additional metadata from the master node when necessary.
1. A method, comprising:
obtaining, by a node that performs one or more data operations, metadata associated with processing at least part of a query submitted to a system via a client node via one or more networks, wherein the metadata is communicated to the node from a master node;
processing, by the node, the at least the part of the query;
determining, by the node, that additional metadata is required in connection with processing the at least the part of the query;
in response to a determination that the additional metadata is required in connection with processing the at least the part of the query, communicating, by the node, a request for the additional metadata, wherein the request for additional metadata is a multicast request to one or more other nodes, wherein the node maintains a list of the other nodes;
after a predetermined period of time:
determining that the one or more other nodes do not have the additional metadata; and
communicating, by the node, a request for the additional metadata to the master node.
2. The method of claim 1 , wherein the metadata is communicated to the node in connection with an indication of the at least the part of the query that is to be processed by the node.
3. The method of claim 2 , wherein the master node communicates a query plan to the node, the query plan comprising the indication of the at least the part of the query that is to be processed by the node.
4. The method of claim 3 , wherein the metadata is comprised in the query plan.
5. The method of claim 3 , wherein the metadata is stored in a database catalog that comprises database objects, and the metadata is associated with corresponding ones of the database objects.
6. The method of claim 5 , wherein the master node obtains the metadata from the database catalog, and the node obtains the metadata in connection with the indication of the at least the part of the query that is to be processed by the node.
7. The method of claim 6 , wherein a separate catalog server session is established between the master node and the database catalog for each query received by the master node.
8. The method of claim 7 , wherein in response to establishing a catalog server session, a snapshot of a metadata catalog corresponding to the database catalog is stored.
9. The method of claim 8 , wherein in response to determining that the one or more other nodes do not have the additional metadata, the additional metadata is obtained from the snapshot of the metadata catalog.
10. The method of claim 8 , wherein during execution of the query, any update to the metadata catalog does not cause an update to the snapshot of the metadata catalog corresponding to the query.
11. The method of claim 1 , wherein in response to a determination that one of the one or more other nodes has the additional metadata, the one of the one or more other nodes communicates the additional metadata to the node.
12. The method of claim 1 , further comprising:
running, by the node, a query execution process in response to receiving a query plan comprising an indication of the at least the part of the query that is to be processed by the node, wherein the node comprises the query execution process that is configured to execute a database query.
13. A node, comprising:
one or more processors configured to:
obtain metadata associated with processing at least part of a query submitted to a system via a client node via one or more networks, wherein the metadata is communicated to the node from a master node;
process the at least the part of the query;
determine that additional metadata is required in connection with processing the at least the part of the query; and
in response to a determination that the additional metadata is required in connection with processing the at least the part of the query, communicating, by the node, a request for the additional metadata, wherein the request for additional metadata is a multicast request to one or more other nodes, wherein the node maintains a list of the other nodes;
one or more memories coupled to the one or more processors, configured to provide the one or more processors with instructions;
after predetermined period of time:
determine that the one or more other nodes do not have the additional metadata; and
communicate, by the node, a request for the additional metadata to the master node.
14. A computer program product, the computer program product being embodied in a non-transitory computer readable storage medium and comprising computer instructions for:
obtaining, by a node that performs one or more data operations, metadata associated with processing at least part of a query submitted to a system via a client node via one or more networks, wherein the metadata is communicated to the node from a master node;
processing, by the node, the at least the part of the query;
determining, by the node, that additional metadata is required in connection with processing the at least the part of the query;
in response to a determination that the additional metadata is required in connection with processing the at least the part of the query, communicating, by the node, a request for the additional metadata, wherein the request for additional metadata is a multicast request to one or more other nodes, wherein the node maintains a list of the other nodes;
after predetermined period of time:
determine that the one or more other nodes do not have the additional metadata; and
communicate, by the node, a request for the additional metadata to the master node.
15. A system, comprising:
a node, comprising:
one or more node processors configured to:
obtain metadata associated with processing at least part of a query submitted to a system via a client node via one or more networks, wherein the metadata is communicated to the node from a master node;
process the at least the part of the query;
determine that additional metadata is required in connection with processing the at least the part of the query; and
in response to a determination that the additional metadata is required in connection with processing the at least the part of the query, communicating, by the node, a request for the additional metadata, wherein the request for additional metadata is a multicast request to one or more other nodes, wherein the node maintains a list of the other nodes; and
one or more node memories coupled to the one or more node processors, configured to provide the one or more node processors with instructions; and
a master node, comprising:
one or more master node processors configured to:
obtain the query from the client node;
develop a query plan based at least in part on the query;
determine the metadata for the node to use in connection with the at least the part of the query;
obtain the metadata from a database catalog that comprises database objects, wherein the metadata is associated with corresponding ones of the database objects; and
provide, to the node, an indication of the at least the part of the query to be processed by the node and the corresponding metadata;
one or more master node memories coupled to the one or more master node processors, configured to provide the one or more master node processors with instructions;
after predetermined period of time:
determine that the one or more other nodes do not have the additional metadata; and
communicate, by the node, a request for the additional metadata to the master node.