Elastic Search geo distance query with MySQL relationships

Question

I would like to make a Elastic Search geo distance query in my MySQL database with relationships. I have one table with the location data and then I have another tables that have relationships with the locations table. I know that NoSQL databases like Elastic Search is not optimised for relationships like this, but is it possible?

This is what my database schema looks like:

CREATE TABLE `locations` (
`id` int(10) unsigned NOT NULL AUTO_INCREMENT,
  `name` varchar(255) COLLATE utf8_unicode_ci NOT NULL,
  `description` varchar(255) COLLATE utf8_unicode_ci NOT NULL,
  `lng` decimal(12,8) NOT NULL,
  `lat` decimal(12,8) NOT NULL,
  `deleted_at` timestamp NULL DEFAULT NULL,
  `created_at` timestamp NULL DEFAULT NULL,
  `updated_at` timestamp NULL DEFAULT NULL,
  PRIMARY KEY (`id`)
) ENGINE=InnoDB AUTO_INCREMENT=26 DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci;

CREATE TABLE `posts` (
`id` int(10) unsigned NOT NULL AUTO_INCREMENT,
  `author` int(10) unsigned NOT NULL,
  `location_id` int(10) unsigned NOT NULL,
  `title` varchar(255) COLLATE utf8_unicode_ci NOT NULL,
  `text` text COLLATE utf8_unicode_ci NOT NULL,
  `deleted_at` timestamp NULL DEFAULT NULL,
  `created_at` timestamp NULL DEFAULT NULL,
  `updated_at` timestamp NULL DEFAULT NULL,
  PRIMARY KEY (`id`),
  KEY `posts_author_foreign` (`author`),
  KEY `posts_location_id_foreign` (`location_id`),
  CONSTRAINT `posts_author_foreign` FOREIGN KEY (`author`) REFERENCES `users` (`id`),
  CONSTRAINT `posts_location_id_foreign` FOREIGN KEY (`location_id`) REFERENCES `locations` (`id`)
) ENGINE=InnoDB AUTO_INCREMENT=174 DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci;

CREATE TABLE `comments` (
`id` int(10) unsigned NOT NULL AUTO_INCREMENT,
  `post_id` int(10) unsigned NOT NULL,
  `author` int(10) unsigned NOT NULL,
  `title` varchar(255) COLLATE utf8_unicode_ci NOT NULL,
  `text` text COLLATE utf8_unicode_ci NOT NULL,
  `deleted_at` timestamp NULL DEFAULT NULL,
  `created_at` timestamp NULL DEFAULT NULL,
  `updated_at` timestamp NULL DEFAULT NULL,
  PRIMARY KEY (`id`),
  KEY `comments_author_foreign` (`author`),
  KEY `comments_post_id_foreign` (`post_id`),
  CONSTRAINT `comments_author_foreign` FOREIGN KEY (`author`) REFERENCES `users` (`id`),
  CONSTRAINT `comments_post_id_foreign` FOREIGN KEY (`post_id`) REFERENCES `posts` (`id`)
) ENGINE=InnoDB AUTO_INCREMENT=238 DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci;

Here's my index mapping(I use the official Elasticsearch client for PHP ):

<?php
return [
    'index' => 'foodie',
    'body' => [
        'mappings' => [
            'locations' => [
                'properties' => [
                    'id' => ['type' => 'string', 'index' => 'not_analyzed'],
                    'name' => ['type' => 'string'],
                    'description' => ['type' => 'string'],
                    'location' => ['type' => 'geo_point'],
                ],
            ],
            'posts' => [
                'properties' => [
                    'id' => ['type' => 'string', 'index' => 'not_analyzed'],
                    'author' => ['type' => 'string', 'index' => 'not_analyzed'],
                    'location_id' => ['type' => 'string', 'index' => 'not_analyzed'],
                    'title' => ['type' => 'string'],
                    'text' => ['type' => 'string'],
                ],
            ],
            'comments' => [
                'properties' => [
                    'id' => ['type' => 'string', 'index' => 'not_analyzed'],
                    'author' => ['type' => 'string', 'index' => 'not_analyzed'],
                    'post_id' => ['type' => 'string', 'index' => 'not_analyzed'],
                    'title' => ['type' => 'string'],
                    'text' => ['type' => 'string'],
                ],
            ]
        ],
        'settings' => [
            'analysis' => [
                'filter' => [
                ],
                'analyzer' => [
                ],
            ],
        ],
    ],
];

I would like to make a query on locations and posts(and comment (= two joins away) as well if that is not too bad for performance) that I can filter and sort by geo distance.

I have tried with a query like this:

[
    'index' => 'index_name',
    'type' => [
        0 => 'posts',
        1 => 'locations',
        2 => 'comments'
    ],
    'body' => [
        'from' => 0,
        'size' => 10,
        'query' => [
            'bool' => [
                'must' => [
                    'multi_match' => [
                        'query' => 'search string',
                        'fields' => [
                            0 => 'title',
                            1 => 'text',
                            2 => 'name',
                            3 => 'description',
                        ],
                        'fuzziness' => 'AUTO',
                        'operator' => 'and',
                    ],
                ],
                'filter' => [
                    'geo_distance' => [
                        'distance' => '100m',
                        'location' => [
                            'lat' => 79.861,
                            'lon' => 107.31,
                        ],
                    ],
                ],
            ],
        ],
    ],
]

It works, but obviously filters out everything except locations that has the location data. How can I include the related posts and possibly even comments into the query?

Thanks!

Answer 1

I didn't find any good solutions to do this, but I solved it by simply adding a location field on both posts and comments and then getting the coordinates of the related location when I push it into the Elasticsearch index.

Maybe not an optimal solution, but it works good and it is very fast as the index is kept completely flat. As long as I make sure the changes in the location coordinates propagate to the related posts and comments I don't see any real issues with this approach except that it's not so clean to store duplicated data in the index.