Overview

Connecting client applications to DB Optimizer (DBO) requires changes to the connection string. Specifically, these are:

• hostname - update the hostname or IP to connect to the DBO Proxy component, rather than directly to the database.
• port - update the port property to 5432 for PostgreSQL and 3306 for MySQL.

When you first connect your applications to DBO, the cache will operate in passthrough mode by default. This allows you to:

• Monitor your database traffic and query patterns
• Analyze potential caching benefits
• Ensure everything is working correctly before enabling active caching

Once you're comfortable with the setup and have reviewed the metrics, you can switch to Auto or Manual mode to start benefiting from DBO's caching capabilities.

Postgres and MySQL

Connection string

In this method, you will update your database connection string in your application code:

1. In the DBO console, navigate to the Databases table.

2. Click on the name of the logical database you want to connect to.

3. In the drawer that opens on the right, locate the Endpoints tab.

4. Copy the connection string and use it in your application.

Testing PostgreSQL Connectivity

CLI Example

psql "host=[Proxy host or IP] \
sslmode=require \
port=9010 \
user='dbuser' \
dbname='dbname'

Cluster setup

When configuring a database cluster (e.g., Amazon Aurora) or using read replicas, it is recommended to create a dedicated endpoint for each instance. This ensures that every database instance hostname has a corresponding DBO endpoint, enabling a smooth and seamless transition to DBO. Example:

In this case, if your application connects to two separate endpoints (e.g., writer-product and reader-product), these endpoints should be updated with the corresponding values listed below. DBO is shared across all members of your cluster, allowing it to observe cache invalidations and traffic from every instance. This improves the overall caching efficiency.

When a request is made to the reader endpoint and results in a cache miss, DBO forwards the query to the appropriate read replica. The same logic applies to the writer endpoint, where cache misses are directed to the primary instance.