SYSTEM AND METHOD FOR IMPROVING A QUERY RESPONSE RATE BY MANAGING A COLUMN-BASED STORE IN A ROW-BASED DATABASE

• US 20160253382A1
• Filed: 02/26/2016
• Published: 09/01/2016
• Est. Priority Date: 02/26/2015
• Status: Abandoned Application

• Current Assignee: Dianomic Systems Incorporated
• Original Assignee: Ori Software Development Ltd
• Inventors: SHADMON, Moshe
• Application Number:
  US15/055,201
• Publication Number:
  US 20160253382A1
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/221 Column-oriented storage; Management thereof
• CPC Class Codes: G06F 16/2453 Query optimisation

Abstract

A system including a shared disk database cluster including one or more database nodes, and at least two storage nodes including a first storage node or nodes with data organized in a row-based format, and a second storage node or nodes with data organized in a column-based format. A method for efficiently searching utilizing a shared disk database cluster including one or more database nodes, and at least two storage nodes including a first storage node or nodes with data organized in a row-based format, and a second storage node or nodes with data organized in a column-based format.

Claims

1. A system comprising:

a shared disk database cluster comprising;

one or more database nodes; and

at least two storage nodes including a first storage node or nodes with data organized in a row-based format, and a second storage node or nodes with data organized in a column-based formant, wherein

each of the storage nodes comprise a memory; and

each of the one or more database nodes comprise a processor configured to;

when a data write operation is received, write a copy of the data to each of the first storage node(s) and the second storage node(s); and

when a data search request is received;

(i) determine whether the first storage node(s) or the second storage node(s) would be more efficient in producing a result for the received data search request;

(ii) transmit the data search request to the first storage node(s) when the first storage node(s) is determined to be more efficient for the request, and to the second storage node(s) when the second storage node(s) is determined to be more efficient for the request; and

(iii) receive the results of the query from the storage node(s) that is determined to be more efficient.

3. The system of claim 1, wherein the copy of data written to each of the storage nodes is the same data.

4. The system of claim 1, wherein the ACID properties are maintained.

8. The system of claim 1, further comprising one or more user terminal devices that are configured to:

receive user input of a search request;

transmit the search request to one of the database node(s);

receive a result for the search request from the database node that the search request was transmitted to; and

display, on a display screen, information regarding the user input based on the result of the search request.

10. The system of claim 1, wherein the determination of whether the first storage node(s) or second storage node(s) are more appropriate is based on comparing previous query response rates.

2. A method for a shared disk database cluster comprising one or more database nodes and at least two storage nodes including a first storage node or nodes with data organized in a row-based format, and a second storage node or nodes with data organized in a column-based formant, the method comprising:

when a data write operation is received by a processor of one of the database nodes, writing a copy of the data to each of the first storage node(s) and the second storage node(s),

when a data search request is received by the processor;

(i) determining whether the first storage node(s) or the second storage node(s) would be more efficient in producing a result for the received data search request; and

(ii) transmitting the data search request to the first storage node(s) when the first storage node(s) is determined to be more efficient for the request, and to the second storage node(s) when the second storage node(s) is determined to be more efficient for the request; and

receiving the results of the query from the storage node(s) that is determined to be more efficient.

5. The method of claim 2, wherein the queries are satisfied from a row-based store or a columnar-store, the row based store including all the data and the column-based store including all or part of the data.

6. The method of claim 5, wherein a query is satisfied by information retrieved from the row based store and the column based store.

7. The method of claim 2, further comprising:

pushing the search request to several storage nodes, said storage nodes searching using a column-organized file,

when data that satisfies the search criteria is found, sending said data to the database node, said database node receiving the data from multiple storage nodes, said query continues to process data using the row based store, said data is returned to the application.

9. A system comprising:

a shared disk database cluster comprising;

one or more database nodes; and

at least two storage nodes including a first storage node or nodes with data organized in a row-based format, and a second storage node or nodes with data organized in a column-based formant, wherein

each of the storage nodes comprise a memory; and

each of the one or more database nodes comprise a processor configured to;

when a data write operation is received, write a copy of the data to each of the first storage node(s) and the second storage node(s); and

when a data search request is received;

(i) determine whether the first storage node(s) or the second storage node(s) is more appropriate for producing a result for the received data search request based on a query response rate;

(ii) transmit the data search request to the first storage node(s) when the first storage node(s) is determined to be more appropriate for the request, and to the second storage node(s) when the second storage node(s) is determined to be more appropriate for the request; and

(iii) receive the results of the query from the storage node(s) that is determined to be more efficient.