Intel® Ethernet Controller - Predictable Load Distribution Using Partial Toeplitz Hash Collections Technology Guide
Last Updated: May 13, 2022
In most cases modern NICs use Toeplitz hash function to distribute packets across the queues with receive side scaling, aka RSS. Due to nature of that distribution it is not possible for the user to select desired queue. Such shortcoming becomes especially noticeable for the systems with dynamic assignment of flow attributes (i.e., parts of hashing tuple such as IP addresses, TCP/UDP ports, ESP SPI, MPLS labels). In this article we present a mechanism for predictable binding of flows to arbitrary queues on the remote receiver or on the local NIC for response packets.
Categories
Categories - Hardware Platforms
Categories - NFVI Hardware
Categories - NFVI OS/Virtualization
Intel Technologies and Platforms - 3rd gen Intel Xeon Scalable processor
Network Location - Core Network
Network Location - Data Center
Network Location - Device Edge
Network Location - Hybrid Cloud
Network Location - On Premises
Network Location - Radio Access Network (RAN)
Network Technologies - 4G and 3G
Network Technologies - 5G
Network Technologies - Cloud Native/NFV
Network Technologies - Containers/Kubernetes
Network Technologies - Multi-Access Edge Computing (MEC)
Network Technologies - Network Edge
Network Technologies - Network Security
Network Technologies - Next Generation Central Office (NGCO)
Network Technologies - SD-WAN/uCPE
Network Technologies - SDN
Network Technologies - Visual Cloud
Network Technologies - vRAN
Optimizations - DPDK
Optimizations - OVS-DPDK
Verticals/Industries - Automotive
Verticals/Industries - Cloud Service Providers
Verticals/Industries - Enterprise
Verticals/Industries - Financial Services
Verticals/Industries - Government
Verticals/Industries - Hospitality
Verticals/Industries - Industrial
Verticals/Industries - Manufacturing
Verticals/Industries - Media and Entertainment
Verticals/Industries - Medical and Healthcare
Verticals/Industries - Retail
Verticals/Industries - System Integration
Verticals/Industries - Telecommunications
Verticals/Industries - Transportation
VNFS and CNFs - Application Delivery Controller (ADC)
VNFS and CNFs - Broadband Remote Access Server (BRAS)
VNFS and CNFs - Content Delivery Networks (CDN)
VNFS and CNFs - Firewall (FW)
VNFS and CNFs - IP Multimedia Subsystems (IMS)
VNFS and CNFs - Load Balancer (LB)
VNFS and CNFs - Session Border Controller (SBC)
VNFS and CNFs - Virtualized Evolved Packet Core (vEPC)
VNFS and CNFs - Virtualized Radio Access Network (vRAN)
VNFS and CNFs - Wide Area Network (WAN) Optimization
Workloads and Use cases - 5G Control Plane and Data Plane
Workloads and Use cases - Artificial Intelligence (AI)
Workloads and Use cases - Augmented/Virtual Reality (AR/VR)
Workloads and Use cases - Cloud Gaming
Workloads and Use cases - Computer Vision
Workloads and Use cases - Connected Cars
Workloads and Use cases - Content Distribution (CDN)
Workloads and Use cases - Emergency Response
Workloads and Use cases - Gaming as a Service
Workloads and Use cases - Media Analytics
Workloads and Use cases - Network Security (Nginx, Next-Gen Firewalls, etc.)
Workloads and Use cases - SD-WAN and uCPE
Workloads and Use cases - Security
Workloads and Use cases - Smart Cities
Workloads and Use cases - Smart Factory
Workloads and Use cases - Smart Retail
Workloads and Use cases - Storage as a Service (SaaS)
Workloads and Use cases - Video Surveillance
Workloads and Use cases - Virtualized Small Cell Gateway
Hide
Related Content
