Cover Image
Market Research Report

Multicore Processor & Switch Survey 2018

Published by Heavy Reading Product code 696431
Published Content info 21 Pages
Delivery time: 1-2 business days
Price
Back to Top
Multicore Processor & Switch Survey 2018
Published: September 11, 2018 Content info: 21 Pages
Description

Integrated multicore processors and switch chips are at the center of networking systems and general-purpose processors, designed for servers, are also widely used for packet processing. Virtual network functions (VNFs) running on server processors are replacing physical systems at both the core and edge of the network. Integrated multicore processors, with standard processor cores and hardware accelerators developed from network processors, provide cost and power efficient solutions for packet processing in virtualized networks. The use of these multicore processors is continually developing with market-place consolidation and many new processor architectures becoming available.

Two key themes coming through in the survey are increasing performance and enhanced programmability. Packet processing at 10 Gbit/s and 100 Gbit/s will be the mainstream with growing demand for 400 Gbit/s. Network processors were developed to bring programmability to the data plane. Integrated multicore processors and general-purpose multicore processors have taken on much of this packet processing and enabled the development of a virtualized network using SDN and NFV.

Until recently, most switch chips implemented a fixed set of switch functions. Programmability was the clear winning feature for switches in the survey. A clear differentiator for several new entrants to the switch chip market has been programmability, and we are now seeing all vendors implement this in their switch chips. The development of high-level languages, such as P4, for packet processing is expected to have a significant impact as the market moves to deliver on the programmable data plane, enabling carriers to define in software the full functionality of the network from controller down to the packet forwarding.

“Multicore Processor & Switch Survey 2018” analyzes the current and projected use of multicore processors and switch chips by telecom and networking equipment manufacturers, based on the results of an exclusive worldwide survey of engineers, designers, product managers and sales/marketing personnel that work for telecom and networking system equipment manufacturers and suppliers. The responses to our survey make it clear that these devices are critical components in most types of networking equipment, from the access edge to the core of the network.

The capacity of switch chips has dramatically increased over the last few years. The first 3.2 Tbit/s switch chips were introduced by Broadcom in 2014, the first 6.4 Tbit/s switch chips were introduced in 2016 and Broadcom announced the Tomahawk 3 12.8 Tbit/s switch chips in December 2017. Innovium introduced a 12.8 Tbit/s switch chip in March 2018 and several other vendors are developing 12.8 Tbit/s devices.

“Multicore Processor & Switch Survey 2018” is published in PDF format.

Table of Contents

Table of Contents

1. EXECUTIVE SUMMARY

  • 1.1. Key Findings
  • 1.2. Companies Covered

2. INTRODUCTION

  • 2.1. Machine Learning Categories
  • 2.2. Why the Resurgence of Interest in AI/ML?

3. POTENTIAL AI/ML USE CASES IN TELECOM

  • 3.1. Network Operations Monitoring & Management
  • 3.2. Predictive Maintenance
  • 3.3. Fraud Mitigation
  • 3.4. Cybersecurity
  • 3.5. Customer Service & Marketing Virtual Digital Assistants
  • 3.6. Intelligent CRM Systems
  • 3.7. CEM

4. CSP ADOPTION OF AI

  • 4.1. TCS Study Suggests Most CSPs Already Using AI/ML in IT/Networking
  • 4.2. TM Forum Study Suggests More Cautious Adoption of AI/ML
  • 4.3. Customer Experience the Key Driver for AI/ML
  • 4.4. AI/ML in Network Management

5. REAL-WORLD CSP EXAMPLES

  • 5.1. AT&T
  • 5.2. COLT
  • 5.3. Deutsche Telekom
  • 5.4. Globe Telecom
  • 5.5. KDDI
  • 5.6. KT
  • 5.7. SK Telecom
  • 5.8. Swisscom
  • 5.9. Telefónica
  • 5.10. Vodafone

6. CHALLENGES OF APPLYING AI/ML TO NETWORKING

  • 6.1. Data That Is Dirty, Unavailable or Difficult to Access
  • 6.2. Lack of Data Science Talent
  • 6.3. Lack of a Clear Question to Answer
  • 6.4. Limitations of Tools

7. ACADEMIC, SDO, CONSORTIA & OS INITIATIVES

  • 7.1. Academia - Knowledge-Defined Networking
  • 7.2. Standards Development Organizations
  • 7.3. Industry Consortium - Telecom Infra Project
  • 7.4. Open Source - Acumos

8. VENDOR PROFILES

  • 8.1. Afiniti
  • 8.2. AIBrain
  • 8.3. Anodot
  • 8.4. Arago
  • 8.5. Aria Networks
  • 8.6. Avaamo
  • 8.7. B.Yond
  • 8.8. Cardinality
  • 8.9. Guavus
  • 8.10. Intent HQ
  • 8.11. IPsoft
  • 8.12. Nuance Communications
  • 8.13. Skymind
  • 8.14. Subtonomy
  • 8.15. Tupl
  • 8.16. Wise Athena

9. CONCLUSIONS

TERMS OF USE

MULTICORE & SWITCHVENDORSCOVERED:

  • Advanced Micro Devices Inc. (Nasdaq: AMD; NYSE: AMD)
  • Ampere ComputingLLC
  • Barefoot Networks Inc.
  • Broadcom Inc. (Nasdaq: AVGO)
  • Centec Networks (Su Zhou) Co. Ltd.
  • HiSilicon Technologies Co. Ltd., a subsidiary of Huawei Technologies Co. Ltd.
  • Intel Corp. (Nasdaq: INTC)
  • Innovium Inc.
  • Marvell Technology Group Ltd. (Nasdaq: MRVL)
  • Mellanox Technologies Ltd.(Nasdaq: MLNX)
  • Nephos (USA) Inc.
  • NXP Semiconductors N.V. (NASDAQ: NXPI)
  • Qualcomm Inc. (Nasdaq: QCOM)
Back to Top