Designing a HTTP-RESTful-API-based infrastructure provisioning system is challenging. Concurrent users, race conditions, erroneous OS responses, quota mechanisms etc. need to be wrapped up under a consistent and responsive CRUD interface. The talk is about a few design patterns and about the "lessons learnt" in building/operating/iterating a large scale API-driven infra-provisioning system.

Sachin Agarwal Bio

In Brief

Python hacker with a day job in the infra/platform cloud space: devops automation, monitoring, capacity planning, data mining logs, virtualization/LXC-docker containerization, software-defined-* (networking/storage/compute).

Formal Stuff


I work as an engineering manager for a infrastructure/platform software development team at TD @ Waterloo. Shameless plug - I am hiring Python developers, especially those with infra/cloud/config management/devops experience - in Waterloo.


  • Storage architect at blackberry - developer of a large scale software-defined-storage all-flash system: Backend for 1000+ node Cassandra cluster(s).
  • Researcher at NEC Labs (Heidelberg, Germany) and Deutsche Telekom Labs (Berlin, Germany), PhD - Boston University.

iCalendar File JSON File