Delaying one or more searches to cause multiple searches to load and evaluate segments together
The performance of a server can be improved by causing multiple searches to load segments of an index and search within them together. When multiple search requests are received and executed within a minimal time difference, the first search can be delayed to allow the second search—and possibly additional searches—to catch up so that the searches do not require loading segments multiple times.
1. A method for optimizing the performance of a server that executes searches of an index, the method comprising:
receiving a first client request comprising a first query for searching the index;
commencing execution of the first query by evaluating the first query against a first segment of the index;
receiving a second client request comprising a second query for searching the index;
commencing execution of the second query after the first query has been evaluated against one or more additional segments;
temporarily delaying the execution of the first query while the second query is evaluated against the first segment and the one or more additional segments; and
once the second query has been evaluated against the first segment and the one or more additional segments, proceeding with the execution of the first query and the second query together by loading subsequent segments of the index and evaluating the first and second queries against the subsequent segments.
2. The method of claim 1 , wherein commencing execution of the first query comprises loading the first segment of the index.
3. The method of claim 1 , wherein evaluating the first query against the one or more additional segment comprises loading the one or more additional segments.
4. The method of claim 1 , wherein commencing the execution of the second query comprises loading the first segment of the index.
5. The method of claim 1 , wherein the first segment and the one or more additional segments are contiguous.
6. The method of claim 1 , wherein the one or more additional segments comprise a plurality of additional segments.
7. The method of claim 1 , wherein loading the subsequent segments of the index and evaluating the first and second queries against the subsequent segments comprises loading each remaining segment of the index and evaluating the first and second queries against each remaining segment.
8. The method of claim 1 , further comprising:
prior to temporarily delaying the execution of the first query, estimating how long the execution of the first query would have to be delayed to allow the execution of the first and second queries to proceed together; and
determining that the estimate is within a minimal time difference.
9. The method of claim 8 , wherein the estimate is based on an average amount of time required to load segments.
10. The method of claim 8 , wherein the estimate is based on how many segments the first query has been evaluated against.
11. The method of claim 1 , further comprising:
receiving one or more additional client requests that each comprise an additional query for searching the index;
commencing execution of the one or more additional queries after the first and second queries have commenced execution; and
temporarily delaying the execution of the first and second queries while the one or more additional queries are executed to thereby allow the execution of the first query, the second query, and the one or more additional queries to proceed together.
12. The method of claim 1 , wherein temporarily delaying the execution of the first query comprises preventing the first query from loading the subsequent segments of the index.
13. The method of claim 1 , wherein the index is stored on network storage.
14. The method of claim 1 , wherein the first and second client requests are received from one or more clients via a REST API.
15. A system comprising:
network storage that stores an index; and
a server that is configured to access the network storage to load segments of the index, the server including an interface for receiving client requests to search the index and a processor for executing the searches, wherein the processor is configured to perform the following method to optimize the performance of the server:
receiving a first client request comprising a first query for searching the index;
commencing execution of the first query by evaluating the first query against a first segment of the index;
receiving a second client request comprising a second query for searching the index;
commencing execution of the second query after the first query has been evaluated against one or more additional segments;
temporarily delaying the execution of the first query while the second query is evaluated against the first segment and the one or more additional segments; and
once the second query has been evaluated against the first segment and the one or more additional segments, proceeding with the execution of the first query and the second query together by loading subsequent segments of the index and evaluating the first and second queries against the subsequent segments.
16. The system of claim 15 , wherein the processor temporarily delays the execution of the first query upon determining that the second query is commenced within a minimal time difference that is calculated at least partially on how many segments the first search has been evaluated against.
17. The system of claim 16 , wherein the minimal time difference is also calculated by identifying an average amount of time required to load a segment.
18. The system of claim 15 , wherein a first subsequent segment is loaded prior to temporarily delaying the execution of the first query such that proceeding with the execution of the first query and the second query together comprises evaluating the first and second queries against the first subsequent segment that was already loaded and then loading a second subsequent segment and evaluating the first and second queries against the second subsequent segment.
19. One or more computer storage media storing computer-executable instructions which when executed by one or more processors implement a method comprising:
commencing execution of a first query by evaluating the first query against a first segment of an index;
commencing execution of a second query after the first query has been evaluated against one or more additional segments;
based on an estimate of how long the execution of the first query would need to be delayed to allow the execution of the first and second queries to proceed together, temporarily delaying the execution of the first query while the second query is evaluated against the first segment and the one or more additional segments; and
once the second query has been evaluated against the first segment and the one or more additional segments, proceeding with the execution of the first query and the second query together by loading subsequent segments of the index and evaluating the first and second queries against the subsequent segments.
20. The computer storage media of claim 19 , wherein the execution of the first query is temporarily delayed in response to determining that the estimate is within a minimal time difference.