Fleetback - Redesign of the database architecture for scaling

About the customer

Fleetback is developing innovative SaaS products for the automotive industry for car dealerships sales and after sales services.

Fleetback helps car dealerships in their digital transformation journey by developing high business value product with a strong focus on quality and user experience. This application aims to connect car dealers with their customers via the use of videos and photos.

The Fleetback platform is made of mobile, web and SmartTV applications for the Sales and After Sales departments of car dealerships.

www.fleetback.com

About the project

Fleetback was a suite of products running in various Cloud SaaS (incl. AWS) and on-prem infrastructure. Everything has now been migrated to AWS Cloud. Fleetback is providing its own services, but it also highly integrated with internal dealership tools. The most common integration Fleetback is providing is with DMS (Dealership management System).

Due to the increase of usage of Fleetback, some performance issues in the database searches have been raised. Impacts were spotted on global application searches and in DMS data as the volume of data has been increasing highly over the days. Prior to the project, all the application data was residing in a MongoDB SaaS. It has been migrated to Amazon Aurora, using the PostgreSQL engine.

It was then requested to provide a more scalable and performant solution matching following requirements:

• 01.

  Improve global database searches performance

• 02.

  Decrease DMS data storage size in PostgreSQL

• 03.

  Improve DMS data insertion performance

• 04.

  Improve search performance inside DMS data

Project Solution

The solution designed and implemented to improve database searches at the best running costs and following best practices:

• 01.

  Add a second PostgreSQL read replica instance

• 02.

  Amazon DynamoDB to store DMS data

• 03.

  Migration of the data in scope from on-prem PostgreSQL to Amazon DynamoDB

• 04.

  DynamoDB schema design to match the most common search filters

• 05.

  Migration of the entire solution from on-prem to AWS Cloud

Project Results

The project has been successfully executed thru a joint effort of Fleetback and ARHS Spikeseed teams and has achieved the following results:

• 01.

  Better separation of external data (DMS) and application data (Fleetback)

• 02.

  PostgreSQL database storage optimization

• 03.

  DMS data search optimization

• 04.

  Better workload balance between 2 database read replica instances

What the customer says...

Fleetback and Spikeseed teams have worked together on achieving these results. They key expertise of Spikeseed in AWS helped us to go faster and improve our knowledge of the AWS services in scope of the project Dimitri Duszynski, Technical Director