Using Elasticsearch to Accelerate Query and Analysis for Relational Databases

Scenario

Overview

Figure 1 Architecture of the Elasticsearch-accelerated relational database solution

1. Service data is stored in the MySQL database.
2. DRS synchronizes data from MySQL databases to the CSS Elasticsearch cluster in real time.
3. Users perform full-text search, Ad Hoc query, and statistics analysis in the Elasticsearch cluster.

Advantages

These advantages make the Elasticsearch cluster an effective supplement to relational databases in processing full-text retrieval and high-concurrency Ad Hoc queries.

Prerequisites

• A MySQL database and an Elasticsearch cluster in security mode are available, and they are in the same VPC and security group.
• Data to be synchronized exists in the MySQL database.

  This document uses the following table structure and initial data as an example.

  1. Create a student information table in MySQL.

     CREATE TABLE `student` (
       `dsc` varchar(100) COLLATE utf8mb4_general_ci DEFAULT NULL,
       `age` smallint unsigned DEFAULT NULL,
       `name` varchar(32) COLLATE utf8mb4_general_ci NOT NULL,
       `id` int unsigned NOT NULL,
       PRIMARY KEY (`id`)
     ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_general_ci;

  2. Insert the initial data of three students into the MySQL database.

     INSERT INTO student (id,name,age,dsc)
     VALUES 
     ('1','Jack Ma Yun','50','Jack Ma Yun is a business magnate, investor and philanthropist.'),
     ('2','will smith','22','also known by his stage name the Fresh Prince, is an actor, rapper, and producer.'),
     ('3','James Francis Cameron','68','the director of avatar');

• An index has been created in the Elasticsearch cluster and matches the table indexes in the MySQL database.
  Run the following command to create an index for the Elasticsearch cluster:

  PUT student
  {
    "settings": {
      "number_of_replicas": 0,
      "number_of_shards": 3
  	},
    "mappings": {
      "properties": {
        "id": {
          "type": "keyword"
  		},
        "name": {
          "type": "short"
  		},
        "age": {
          "type": "short"
  		},
        "desc": {
          "type": "text"
  		}
      }
    }
  }

  Configure number_of_shards and number_of_replicas as needed.

Procedure

1. Use DRS to synchronize MySQL data to CSS in real time. For details, see From MySQL to CSS/ES.
   In this example, configure the parameters by following the suggestions in Table 1.

   Table 1 Synchronization parameters

   Module	Parameter	Suggestion
   Create Synchronization Instance > Synchronize Instance Details	Network Type	Select VPC.
   	Source DB Instance	Select the RDS for MySQL instance to be synchronized, that is, the MySQL database that stores service data.
   	Synchronization Instance Subnet	Select the subnet where the synchronization instance is located. You are advised to select the subnet where the database instance and the Elasticsearch cluster are located.
   Configure Source and Destination Databases > Destination Database	VPC	Select the same VPC as the Elasticsearch cluster.
   	Subnet	Select the same subnet as the Elasticsearch cluster.
   	IP Address or Domain Name	Enter the IP address of the Elasticsearch cluster. For details, see Obtaining the IP address of a CSS cluster.
   	Database Username	Enter the administrator username (admin) and password of the Elasticsearch cluster.
   	Database Password	Enter the administrator password of the Elasticsearch cluster.
   	Encryption Certificate	Select the security certificate of the Elasticsearch cluster. If SSL Connection is not enabled, you do not need to select any certificate. For details, see Obtaining the security certificate of a CSS cluster.
   Set Synchronization Task	Flow Control	Select No.
   	Synchronization Object Type	Deselect Table structure, because the indexes matching MySQL tables have been created in the Elasticsearch cluster.
   	Synchronization Object	Select Tables. Select the database and table name corresponding to the Elasticsearch cluster. NOTE: Ensure the type name in the configuration item is the same as the index name, that is, _doc.
   Process Data	-	Click Next.

   After the synchronization task is started, wait until the Status of the task changes from Full synchronization to Incremental, indicating real-time synchronization has started.

2. Check the synchronization status of the database.
   1. Verify full data synchronization.

      Run the following command in Kibana of the Elasticsearch cluster to check whether full data has been synchronized to CSS:

      GET student/_search

   2. Insert new data in the source cluster and check whether the data is synchronized to Elasticsearch.
      For example, insert a record whose id is 4 in the source cluster.

      INSERT INTO student (id,name,age,dsc)
      VALUES 
      ('4','Bill Gates','50','Gates III is a business magnate, software developer, investor, author, and philanthropist.')

      Run the following command in Kibana of the Elasticsearch cluster to check whether new data has been synchronized to CSS:

      GET student/_search

   3. Update data in the source cluster and check whether the data is synchronized to Elasticsearch.
      For example, in the record whose id is 4, change the value of age from 50 to 55.

      UPDATE student set age='55' WHERE id=4;

      Run the following command in Kibana of the Elasticsearch cluster to check whether the data is updated in CSS:

      GET student/_search

   4. Delete data from the source cluster and check whether the data is deleted synchronously from Elasticsearch.
      For example, delete the record whose id is 4.

      DELETE FROM student WHERE id=4;

      Run the following command in Kibana of the Elasticsearch cluster to check whether the data is deleted synchronously from CSS:

      GET student/_search

3. Verify the full-text search capability of the database.

   For example, run the following command to query the data that contains avatar in dsc in the Elasticsearch cluster:

   GET student/_search
   {
     "query": {
       "match": {
         "dsc": "avatar"
       }
     }
   }

4. Verify the ad hoc query capability of the database.

   For example, query philanthropist whose age is greater than 40 in the Elasticsearch cluster.

   GET student/_search
   {
     "query": {
       "bool": {
         "must": [
           {
             "match": {
               "dsc": "philanthropist"
             }
           },
           {
             "range": {
               "age": {
                 "gte": 40
               }
             }
           }
         ]
       }
     }
   }

5. Verify the statistical analysis capability of the database.

   For example, collect statistics on the age distributions of all users in the Elasticsearch cluster.

   GET student/_search
   {
     "size": 0, 
     "query": {
       "match_all": {}
     },
     "aggs": {
       "age_count": {
         "terms": {
           "field": "age",
           "size": 10
         }
       }
     }
   }

Other Operations

• Obtaining the IP address of a CSS cluster
  1. In the navigation pane on the left, choose Clusters.
  2. In the cluster list, obtain the IP address of a CSS cluster from the Private Network Address column. Generally, the IP address format is <host>:<port> or <host>:<port>,<host>:<port>.

     If the cluster has only one node, the IP address and port number of only one node are displayed, for example, 10.62.179.32:9200. If the cluster has multiple nodes, the IP addresses and port numbers of all nodes are displayed, for example, 10.62.179.32:9200,10.62.179.33:9200.

• Obtaining the security certificate of a CSS cluster
  1. Log in to the CSS management console.
  2. In the navigation pane, choose Clusters. The cluster list is displayed.
  3. Click a cluster name to go to the cluster details page.
  4. In the Configuration area, click Download Certificate next to HTTPS Access.
     Figure 2 Download Certificate