My first blog: DPUs for Datacenters

The Data Center is continuously evolving. Traditionally data centers were on monolithic software packages. The next generation was the virtualization-driven software-defined data centers [SDDC]. In recent years, with the advent of Cloud Native, the applications are disaggregated into smaller microservices that run on Containers and are managed with Kubernetes. The rise of Containers and Kubernetes puts the x86 processor under tremendous pressure. The CPU has to deal with everything from running the operating system, applications, network traffic, storage I/O, security and more. For example, the hypervisors themselves consume about 30% of those CPU cores which impacts the Container applications.

Thus, to deal with the raising need for infrastructure computation, it led to a new breed of Infrastructure Processors called Data Processing Unit [DPU]. DPU is a new class of processor of Datacenter Infrastructure processing. DPUs can be used to offload infrastructure services like software-defined networking, Packet Forwarding, Overlay support, basic TCP functions, Flow Tables, Security functions and more from the host CPU and accelerate them in hardware by leveraging the purpose-built and programmable hardware accelerators within the DPU. Henceforth, Datacenter CPUs are freed up from the infrastructure tasks and the CPUs can run only the Container specific workloads which provides better business outcomes overall!