D4C: Leveraging Delta Encodings for Faster and Lighter Container Image Updating.

Day 1 | 18:10 | 00:20 | UD2.218A | Naoki Matsumoto


Note: I'm reworking this at the moment, some things won't work.

The stream isn't available yet! Check back at 18:10.

Container images sometimes take tens to hundreds of megabytes in size and require many network resources to be transferred. Especially in low-bandwidth network environments like edge computing, frequent image updating will be difficult and increase the network cost. In this talk, I will present "D4C", a software that provides faster and lighter image updating with delta encodings. Delta encodings reduce data size, but generating and applying deltas is a time-consuming operation. D4C utilizes some techniques to reduce them including merge-based delta management and applying deltas lazily. D4C reduces the data size to update images by up to 95% and provides about 10x faster updating in low-bandwidth networks than the state-of-the-art approach. Benchmarks with PostgreSQL and PyTorch showed that their performance degradation caused by D4C is negligible.