Developing Data Processing Pipelines for Massive Sky Surveys - Lessons Learned from SkyMapper

The SkyMapper Survey led by RSAA at ANU started in 2014 to map the Southern sky in six spectral passbands. It will identify over 100 million celestial objects and be the first deep and detailed digital resource covering the entire Southern sky. Here we report on lessons learned from the ongoing effort of developing its Science Data Pipeline. How reliably can the required functionality be anticipated? What issues are posed by dependencies? And, is a supercomputer the best platform?