Holistic optimization by prefetching query results

Abstract

In this paper we address the problem of optimizing performance of database/web-service backed applications by means of automatically prefetching query results. Prefetching has been performed in earlier work based on predicting query access patterns; however such prediction is often of limited value, and can perform unnecessary prefetches. There has been some earlier work on program analysis and rewriting to automatically insert prefetch requests; however, such work has been restricted to rewriting of single procedures. In many cases, the query is in a procedure which does not offer much scope for prefetching within the procedure; in contrast, our approach can perform prefetching in a calling procedure, even when the actual query is in a called procedure, thereby greatly improving the benefits due to prefetching. Our approach does not perform any intrusive changes to the source code, and places prefetch instructions at the earliest possible points while avoiding wasteful prefetches. We have incorporated our techniques into a tool for holistic optimization called DBridge, to prefetch query results in Java programs that use JDBC. Our tool can be easily extended to handle Hibernate API calls as well as Web service requests. Our experiments on several real world applications demonstrate the applicability and significant performance gains due to our techniques.

A program optimization for automatic database result caching

“…Query prefetching (Ramachandra and Sudarshan 2012) relies on program analysis and transformation to take a program written in a natural style and replace it with one that executes several queries at once, reducing the number of database round trips. The Sloth system (Cheung et al 2014) adopts a more dynamic approach.…”

Towards unified ad-hoc data processing

“…Guravannavar et al [13] and Cheung et al [3] proposed different methods to reduce the overhead incurred by the communication between the applications and databases, i.e., rewrite the iteratively invoked queries into a batch form [13], and partition database programs into separate parts and execute some parts in the databases [3]. Chavan et al [2] and Ramachandra et al [23] studied the transformation methods to reduce the program latency by prefetching the query results.…”

Sloth

“…This example shows how Sloth is able to perform much more batching than either the existing "lazy" fetching mode of Hibernate or prior work using static analysis [23]. Hibernate's lazy fetching mode would have to evaluate the results of the database-accessing statements such as getVisitsByPatient(p) on Line 14 as its results are needed by the filtering operation, leaving no opportunity to batch.…”

“…Both of these techniques have been explored in prior research. Latency hiding, which generally takes the form of asynchronously "prefetching" query results so that they are available when needed by the program, was explored by Ramachandra et al [23], where they employed static analysis to identify queries that will be executed unconditionally by a piece of code. The compiler can then transform the code so that these queries are issued as soon as their query parameters are computed, and before their results are needed.…”