Zissou: Novel Datacenter, Server, and Software Designs using Liquid Immersion Cooling
Chip power has been steadily increasing with the end of Dennard scaling, requiring ever larger cooling infrastructures and server footprints. Motivated by this trend, in project Zissou, we are exploring the use of 2-phase liquid immersion cooling for hyperscale public clouds. Zissou will provide a multi-fold improvement in our IT cooling capability, and in turn unlock innovation across datacenter, server, and software stacks. For example, by alleviating thermal constraints, Zissou will enable us to densely pack components and servers, improve the performance of disaggregated-resource architectures, equip servers with high-powered chips and interconnects, and aggressively overclock components. We also expect that it will lower component failure rates, through lower and stable operating temperatures combined with inert environmental conditions.
Zissou unlocks interesting opportunities in our vision towards software-defined datacenters – enabling intelligent failure management, managing the tradeoff between performance, power, and reliability in component overclocking. In this talk, I will introduce Zissou, the main areas we are exploring, and some of our initial results.
Check out https://aka.ms/JoinTeamZissou for a list of positions we are currently hiring.
If you want to learn about cutting edge innovation happening at the intersection of datacenter, hardware systems and software – you should join this talk.
Brijesh Warrier is a Partner Engineering Manager in Microsoft’s Azure organization where he is leading an initiative around software-defined datacenter, helping leverage software and ML to efficiently match workload requirements with Data center and IT infrastructure. He has been with Microsoft for 16 years, with stints at different parts of the company across Windows, Server infrastructure and Azure. He is one of the early members of the core Azure team. He has delivered innovative solutions around predictive hardware failure and mitigation, system for safely oversubscribing Data center power, and Machine Learning for Systems solution. He has an MS degree from University of Cincinnati and a BTech from National Institute of Technology, Allahabad India.