Home > Communication Service Provider Solutions > Telecom Technical White Papers > Sizing a Bare Metal Kubernetes Telco Cloud > DPDK
Data Plane Development Kit (DPDK) is a set of libraries and drivers that enables high-performance packet processing in software applications. A critical requirement of DPDK is dedicated CPU cores to achieve optimal performance.
DPDK uses a “poll mode driver” technique where the software application controls the network interface card (NIC) and bypasses the kernel networking stack. This technique enables faster packet processing and reduced latency. To achieve this performance, DPDK requires dedicated CPU cores that are exclusively allocated to handle the packet processing tasks.
Dedicating CPU cores to DPDK ensures that the packet processing workload is isolated from other tasks and processes running on the system, preventing interference and resource contention. With dedicated CPU cores, DPDK can fully use the available processing power and achieve higher throughput and lower latency.
It is essential to consider the specific requirements of your DPDK-based applications when planning the deployment of your Telco cloud infrastructure. Allocating the appropriate number of dedicated CPU cores to DPDK tasks will help ensure optimal performance and efficient packet processing, ultimately enhancing your network applications' overall performance and responsiveness.
Note: For information about DPDK CPU pinning in OpenShift that is valid for all Kubernetes flavors, see https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/10/html/network_functions_virtualization_planning_guide/assembly_ovsdpdk_parameters.