90% cost reduction in data pipeline management
Industry: E-commerce · Size: 500 employees · Previous stack: Fivetran + AWS Glue + S3 · Sources: 45 connections · Migration: 2 weeks
An $80M e-commerce company replaced Fivetran, Glue jobs, and two engineers' worth of maintenance with one Skippr runner.
| Before | After | |
|---|---|---|
| Tooling | ~$15k/mo (Fivetran + Glue + S3) | Skippr runner on your host; Cloud usage as vCPU time, memory time, bytes stored, and network bytes |
| Breakages | 2–3 per week | 0 in 6 months |
| Dedup layer | Custom SQL after every sync | Removed (exactly-once delivery) |
Problem
Forty-five Fivetran connectors plus Glue jobs for transforms Glue could not cover. Two engineers spent most of their time on schema-drift breakages and at-least-once deduplication. One engineer gave notice; hiring would take months.
Solution
Week one: configure Skippr for the same 45 sources, including CDC where it replaced database connectors. Week two: parallel-run validation (row counts and spot checks). Fivetran and Glue were decommissioned. Schema evolution absorbed type and field changes; dead letters caught malformed records without failing the pipeline.
Outcome
- One engineer moved off pipeline maintenance onto analytics.
- The departing role was not backfilled for ingest.
- Migration completed in two weeks with validated row counts.
See Compare ELT and ELT features. More stories: Customers.