ANT-Man: Towards Agile Power Management in the Microservice Era
Accelerators, FPGA, and GPUs
Reliability and Resiliency
TimeWednesday, 18 November 20204pm - 4:30pm EDT
DescriptionThe emerging trend of decomposing cloud applications into microservices has raised new questions about managing the performance/power trade-off of a datacenter at microsecond-scale. We introduce ANT-Man, an Agile, Native and Transparent power Management framework that can exploit fine-grained microservice variability for system efficiency. To achieve this, ANT-Man abstracts away two major sources of latency overhead in traditional hierarchical power management frameworks. First, ANT-Man proposes an agile power budgeting scheme for reducing power coordination latency at the data center level. It can proactively determine the power budget tailored to each individual microservice. Second, ANT-Man proposes a native and transparent power control scheme to overcome the power configuration latency for each microservice. It enables super-fast power budget enforcement with nanosecond-scale performance scaling. Extensive experiments on our prototyped system show that ANT-Man could slash power consumption by 7.8-43.5%, while reducing the 95th tail latency by 9.7-12.5%, compared to existing techniques.