view release on metacpan or  search on metacpan

lib/BenchmarkAnything/Storage/Backend/SQL.pm  view on Meta::CPAN

=head3 json_false

Auxiliary function for Elasticsearch JSON data.

=head2 Elasticsearch support

=head3 Config

You can pass through a config entry for an external search engine
(currently only Elasticsearch) to the constructor:

    my $or_bench = BenchmarkAnything::Storage::Backend::SQL->new({
        dbh    => $or_dbh,
        searchengine => {
          elasticsearch =>
            #
            # which index/type to use
            index => "myapp",
            type  => "benchmarkanything",
            #
            # queries use the searchengine
            enable_query => 1,
            #
            # should each single added value be stored immediately
            # (maybe there is another bulk sync mechanism)
            index_single_added_values_immediately => 1,
            #
            # which nodes to use
            nodes => [ 'localhost:9200' ],
            #
            # (OPTIONAL)
            # Your additional application specific mappings, used when
            # index is created.
            #
            # WARNING: You are allowed to overwrite the pre-defined
            # defaults for the built-in fields (NAME, VALUE, VALUE_ID,
            # UNIT, CREATED) as it is a RIGHT_PRECEDENT hash merge with
            # your additional_mappings on the right side. So if you
            # touch the internal fields you better know what you are
            # doing.
            additional_mappings => {
                # static key <properties>
                properties => {
                    # field
                    tapper_report => {
                        type => long,
                    },
                    tapper_testrun => {
                        type => long,
                    },
                    tapper_testplan => {
                        type => long,
                    },
                },
            },
        },
    });

With such a config and an already set up Elasticsearch you can use the
lib as usual but it is handling all things transparently behind the
scenes to index and query the data with Elasticsearch. The relational
SQL storage is still used as the primary storage.

=head3 Index

When C<index_single_added_values_immediately> is set, every single
added entry is fetched right after insert (to get all transformations
and added metadata) and sent to elasticsearch for index.

Please note, this immediate indexing adds an overhead to insert
time. You could as well switch-off this setting and take care of
indexing the data at another time. Then again, for instance the
C<::Frontend::HTTP> already takes care of bulk-adding new data
asynchronously, so the overhead should be hidden in there, so just
switch-on the feature and don't worry too much.

=head3 Search

When C<enable_query> is set, the BenchmarkAnything queries are
transformed into corresponding Elasticsearch queries, sent to
Elastisearch, and the result is taken directly from its answers.

=head1 AUTHOR

Roberto Schaefer <schaefr@amazon.com>

=head1 COPYRIGHT AND LICENSE

This software is Copyright (c) 2020 by Amazon.com, Inc. or its affiliates.

This is free software, licensed under:

  The (two-clause) FreeBSD License

=cut