Accelerator Management g-API’s

Slides:



Advertisements
Similar presentations
Zhipeng (Howard) Huang
Advertisements

Keith Wiles DPACC vNF Overview and Proposed methods Keith Wiles – v0.5.
Accelerating the Path to the Guest
Active Messages: a Mechanism for Integrated Communication and Computation von Eicken et. al. Brian Kazian CS258 Spring 2008.
Chapter 13 Embedded Systems
DPACC vNF Overview and Proposed methods Keith Wiles – v0.5.
A. Frank - P. Weisberg Operating Systems Structure of Operating Systems.
dpacc framework discussion data plane
OPNFV Platform Overview
Microkernels, virtualization, exokernels Tutorial 1 – CSC469.
1 Design and Performance of a Web Server Accelerator Eric Levy-Abegnoli, Arun Iyengar, Junehwa Song, and Daniel Dias INFOCOM ‘99.
Design and Implementation of a Linux SCSI Target for Storage Area Networks Ashish A. PalekarAnshul Chaddha, Trebia Networks Narendran Ganapathy, 33 Nagog.
Silicon Graphics, Inc. Re-Configurable Application Specific Computing (RASC/FPGA) David Alexander Director of Engineering.
1 Hardware Security Mechanisms Krste Asanovic U.C. Berkeley August 20, 2009.
Seoul National University
Memory/Storage Architecture Lab 1 Virtualization History of Computing = History of Virtualization  e.g., process abstraction, virtual memory, cache memory,
Architecture Support for OS CSCI 444/544 Operating Systems Fall 2008.
Virtualization: Not Just For Servers Hollis Blanchard PowerPC kernel hacker.
Tizen v2.3 Emulation on PC.
 Virtual machine systems: simulators for multiple copies of a machine on itself.  Virtual machine (VM): the simulated machine.  Virtual machine monitor.
Disco : Running commodity operating system on scalable multiprocessor Edouard et al. Presented by Vidhya Sivasankaran.
CS 346 – Chapter 2 OS services –OS user interface –System calls –System programs How to make an OS –Implementation –Structure –Virtual machines Commitment.
PANA Implementation in Open Diameter Victor Fajardo.
Intel Research & Development ETA: Experience with an IA processor as a Packet Processing Engine HP Labs Computer Systems Colloquium August 2003 Greg Regnier.
Proposed B-release planning for dpacc documentation (for discussion) 9/16/2015 1December 7, 2015 OPNFV Introduction.
A. Frank - P. Weisberg Operating Systems Structure of Operating Systems.
Full and Para Virtualization
PDAC-10 Middleware Solutions for Data- Intensive (Scientific) Computing on Clouds Gagan Agrawal Ohio State University (Joint Work with Tekin Bicer, David.
Operating-System Structures
Virtio-IPsec-LA PoC Implementation
DPACC Management Aspects
11 MANAGING DEVICES AND PERIPHERALS Chapter 4. Chapter 4: Managing Devices and Peripherals2 CHAPTER OVERVIEW  Implement, manage, and troubleshoot input.
Figure A: From Openstack Nomad. Figure B: From Gap on OpenStack ① ① ④ ④.
DPACC Metadata Update Discussion Lingli Deng 2016/05/05.
DPACC Metadata Revised 2016/4/6. Table of Contents Motivation Information Elements Data representation Convergence discussion for IFA004.
Opnfv Summit 2016 (Berlin) DPACC and DPDK solving NFV acceleration
Virtualization.
New Approach to OVS Datapath Performance
NFV Compute Acceleration APIs and Evaluation
BESS: A Virtual Switch Tailored for NFV
DPDK API and Virtual Infrastructure
Kay Ousterhout, Christopher Canel, Sylvia Ratnasamy, Scott Shenker
Tools and Services Workshop Overview of Atmosphere
OPNFV: Support for HA Guest APIs: Introduction
Nov, 2015 Howard Huang, Huawei Julien Zhang, ZTE
Virtio Inline Accelerator
DPACC Management Aspects
PANA Implementation in Open Diameter
Virtio Keith Wiles July 11, 2016.
Integrating DPDK/SPDK with storage application
rte_security: A new crypto-offload framework in DPDK
Reprogrammable packet processing pipeline
Optimizing stencil code for FPGA
EE 4xx: Computer Architecture and Performance Programming
Mixed Up Multiplication Challenge
Update Summary of DPACC docs
A very basic introduction
Latest Update on Gap Analysis of Openstack for DPACC
Flow Processing for Fast Path & Inline Acceleration
DPACC API Guidelines 2019/10/12.
Latest Update on Gap Analysis of Openstack for DPACC
Update Summary of DPACC docs
Latest Update DPACC Use-cases
Virtio-ipsec F.F. Ozog (6WIND) v1 (2015/05/29).
Platform Performance Acceleration
Latest Update DPACC Use-cases
Figure 3-2 VIM-NFVI acceleration management architecture
Latest Update DPACC Architecture
TOSCA Orchestration Paradigm
Presentation transcript:

Accelerator Management g-API’s S.No IPSec – Look Aside Accelerator PDCP – Look Aside Accelerator 1 g_ipsec_la_get_api_version g_pdcp_la_get_api_version 2 g_ipsec_la_open g_pdcp_la_open 3 g_ipsec_la_close g_pdcp_la_close 4 g_ipsec_la_group_delete 5 g_ipsec_la_group_create 6 g_ipsec_la_get_available_list 7 g_ipsec_la_get_active_list 27/10/2019

How to common gAPI (IPSec/PDCP etc.,) The below challenges needs to be discussed to make the Management API’s in common Management Abstraction layer should be super set of all the accelerators. Managing the multiple api version of different lookaside accelerators. Performance overhead on the system, where only one accelerators is used. Multiple instance management for the supported/unsupported accelerators Managing the similarity between the uspace/kspace applications Growing API set for addition of new accelerators All Application has to be changed for the changes in the API parameters VNF Application 1 VNF Application 1 g-API Management Abstraction layer g-API - control g-API - data legacy-API g-API - control g-API - data legacy-API Acceleration Core (AC) e.g. DPDK, ODP or another acceleration implementation s-API Acceleration Core (AC) e.g. DPDK, ODP or another acceleration implementation s-API Software Acceleration Layer (SAL) Buffer and memory mgnt, rings/queues, ingress/egress scheduling, tasks, pipeline, … g-drivers for (paravirtualized) SW/HW-funcs Buffer and memory mgnt, rings/queues, ingress/egress scheduling, tasks, pipeline, … g-drivers for (paravirtualized) SW/HW-funcs sio + VirtIO hio sio + VirtIO hio This view is attempting to how the DPACC layer and configuration can be designed by the developer to give his application the best accelerated performance. In this picture having a software acceleration layer in the host would provide the best performance and flexibility for a VNF application.