Latest Content

Exploring Google Cloud networking enhancements for generative AI applications
Jun 26, 2024
Article

Since joining Google Cloud in 2015 as Uber TL, I conceived of and drove the modernization of GCP application networking by embracing emerging cloud native technologies using Envoy proxy with Traffic Director as open service mesh and universal control plane that powers all Google Cloud application load balancing products. For large deployments and latency critical services I defined industry first gRPC proxyless mesh that improves on the novel concept of service mesh by removing proxy to simplify service mesh operations.

I also drove convergence of parts of Google Cloud infrastructure stack to bring cohesiveness of networking and autoscaling features parity for VM and container based and serverless workloads.

Before that since 2006, as TL of Global Service Load Balancer (GSLB) for Google services, I designed and developed nextgen GSLB that powers load balancing for all Google properties for internet and service-service traffic, while optimizing for latency, workload capacity and health. I oversaw adoption with 1000X growth in the number of services, while ensuring 5+ nines of availability.

Before Google, I worked at Extreme Networks, Cosine Communications and HolonTech, focusing on designing and building highly reliable distributed network control systems in embedded systems.

1
article