Prior Situation / Scenario:
- A leading North American Telecommunications company operating in over 20 countries across LATAM and the Caribbean under different brands.
- They provide several communications and entertainment services to residential and business customers including video, broadband internet, telephone, and mobile services.
- Company deployed a SAS Customer Engagement marketing platform (MA and RTMD), SAS Viya Visual Analytics, and a SAS Data Integration Studio.
- ETL jobs were running on SAS Cluster, based on SAS Data integration, its main database engine being an RDS Oracle.
Client Challenges:
- Increased operational costs based on Oracle DB.
- SAS dedicated instances.
- Cost’s expected growth.
Strata Solution/ Key Enablers:
- Migration and re-factoring from SAS Code and Data Integration jobs to pySpark AWS Glue jobs, orchestrated with AWS Step Functions and Apache Airflow.
- Data Lake is now based on AWS Glue catalog composed of parquet files, presented by Athena connector. Dashboards and analytics are being done on Quicksight.
- Reduction on EC2 instances due to re-clustering of High Availability SAS RTDM nodes. Reduction on RDS instance.
Outcome:
- Substantial Cost Reduction.
- Refactoring and Data Lake migration from SAS to AWS Serverless Glue / pyspark jobs.
- Streamlined operation and simplified pipeline tracking with Airflow.
Results:
After the migration of the SAS Data platform to AWS jobs and the orchestration with Apache Airflow, we were able to reduce database costs and reduce instances costs by a total of 30% compared to prior expenses.