Image Processing for Dropshipping Platform with Serverless AWS Lambda
Challenge
The client's dropshipping platform processed product imports containing thousands of products and multiple images per item. Image downloading, WebP conversion, and database operations overloaded the main server and AWS infrastructure. Imports of 4,000–5,000+ products caused 10–20 minutes of downtime and image-processing delays of 2–3 days, sometimes extending to several days.
The client's dropshipping platform processed product imports containing thousands of products and multiple images per item. Image downloading, WebP conversion, and database operations overloaded the main server and AWS infrastructure. Imports of 4,000–5,000+ products caused 10–20 minutes of downtime and image-processing delays of 2–3 days, sometimes extending to several days.
Solution
Radixweb migrated image processing from 10 EC2 instances to AWS Lambda, introduced Amazon SQS for controlled background processing, automated image imports through a two-hour scheduled workflow, stored processed images in Amazon S3, and increased Amazon RDS IOPS capacity to 5,000. This separated image-processing workloads from the core application and reduced server dependency.
Radixweb migrated image processing from 10 EC2 instances to AWS Lambda, introduced Amazon SQS for controlled background processing, automated image imports through a two-hour scheduled workflow, stored processed images in Amazon S3, and increased Amazon RDS IOPS capacity to 5,000. This separated image-processing workloads from the core application and reduced server dependency.
Results
Image processing improved from 2–5 days to approximately 4–5 hours after product imports. The platform maintained availability during heavy imports of 5,000–8,000+ products, eliminating the previously reported 10–20 minutes of downtime. Stress testing processed approximately 100,000 images within 4–5 hours, validating scalability for future catalog expansion.
Image processing improved from 2–5 days to approximately 4–5 hours after product imports. The platform maintained availability during heavy imports of 5,000–8,000+ products, eliminating the previously reported 10–20 minutes of downtime. Stress testing processed approximately 100,000 images within 4–5 hours, validating scalability for future catalog expansion.