RT Embedded Virtualization · HUAWEI TECHNOLOGIES CO., LTD. 35pt 32pt ) :18pt Page 29 OpenHuawei.org R&D Lab • Web service accessible by everyone at Huawei • via http(s) using

HUAWEI TECHNOLOGIES CO., LTD.

35pt

32pt

) :18pt

OpenHuawei.org

Linux Foundation Collaboration Summit April 5th, 2012, San Francisco, California

Sven-Thorsten Dietrich Huawei Technologies, US Central Research Center

www.openhuawei.org

HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential

35pt

32pt

) :18pt

Linux Adoption: General History

Company Overview

Conclusions

Linux Adoption at Huawei

OpenHuawei.org: Fast-track to upstream

Linux Challenges at Huawei

OpenHuawei.org: Projects

Page 2


35pt

32pt

) :18pt

• Fast-growing Technology Company in Shenzhen, CN

• 62K(+) Engineering organization head count

• Multi-pronged Open Source strategy

• Contributor to Apache Hadoop

• Distributions from Canonical, Suse, RedHat, WindRiver

• Ambitious expansion objective into broader IT market

• Linux seen as strategic technology

• Huawei seeking sophisticated Linux expertise

Page 3

Huawei Company Overview


35pt

32pt

) :18pt

2007 2008 2009

2010 2011

(Unaudited) 2012 (Target)

12.6 (CNY 92.2 bln)

18 (CNY123.1 bln)

21.5 (CNY146.6 bln)

27.6 (CNY182.5 bln)

32.4 (CNY203.9 bln)

38.7 (CNY243.4 bln)

Huawei releases its Annual Report with consolidated financial statements audited by

KPMG

Revenue (USD in billion)

Sustainable Growth

Page 4


35pt

32pt

) :18pt

Continuous Innovation Investment

USD 3.76 billion in 2011

62,000 R&D employees

23 R&D centers, 6 in Europe and 5 in NA

34 Joint Innovation Centers

23,522 patents, 90%+ invention patents

2007 2008

2009 2010 2011

9.4% 9.1% 9.7%

9.7%

11.6% % of revenue R&D

spending

• Huawei’s innovation engine

• Navigating the company’s

future through innovation

• Driving the industry and

society forward

Central R&D

• Cloud Computing will shape

future network development

• Heavy investment in R&D

• Develop datacenter solutions

based on Cloud Computing

Cloud

Computing

Page 5


35pt

32pt

) :18pt

Page 6

History: Gross Profit vs. R&D Budgets

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

90.00%

RHT ORCL IBM INTC AMD ARM AAPL QCOM BRCM CAVM CSCO CNXT ZNGA MRVL GOOG NOK ERIC MENT MSFT

R&D spending per Dollar of Gross Profit

Source: NASDAQ

2008

2009

2010

2011


35pt

32pt

) :18pt

Linux Adoption: General

Company Overview

Conclusions





Page 7


35pt

32pt

) :18pt

• GPL impact on legacy proprietary R&D processes?

•Compliance

• IP segregation

•Maintenance

•Security

Page 8

Linux Adoption History: Risk vs. Benefit?


35pt

32pt

) :18pt

• Can we trust independent developers not under our

direct control?

• Buy or Build?

• Supply chain conflict of interest?

• Vendor technology road-map?

• Feature integration and availability?

• Maintenance: Farm out or in-house?

• Support quality vs. response time?

Page 9

Linux Adoption History: Risk vs. Benefit?


35pt

32pt

) :18pt

• Early Linux adopters displacing incumbents

•Linux outcompetes proprietary platforms

• Proprietary SW market share and profit pressure

•Niches remain: extreme performance / reliability

• Medical

• Nuclear

• Military

• Proprietary OS generally relegated to history textbooks

Page 10

Linux Adoption History: Competition


35pt

32pt

) :18pt

• Software R&D cost factors

• HW arch, platform, driver support

• Emerging technologies

• Quality in bugs/loc

• Up-front and short-term cost to market

• Time to market

• Royalty contributions

• Long-term and life-time COGS: variable • Maintenance & Features

Page 11

Linux Adoption History: Competition


35pt

32pt

) :18pt

• Bottom line drives Linux adoption

• Benefits outweigh the burdens

• Short-term R&D investment vs. product complexity

• Long-term costs:

• often poorly understood

• internal processes

• code-push-back

• internal collaboration

• implementation of industry best-practices

• Code re-use

Page 12

History: Adoption Bottom-Line driven


35pt

32pt

) :18pt

• Pace of Linux development

• Product complexity vs. R&D effort

•Feature base vs. differentiation

•R&D budget vs. market-share vs. max ROI

• Selection by efficiency of adaptation

•R&D budget, product cost, features, profit margin

• Proprietary SW features vs. Linux

• improve R&D efficiency

• increase R&D budget

• Market consensus: Linux competitive

Page 13

Linux Adoption History: R&D budget


35pt

32pt

) :18pt

• Linux engineering (Labor) is not free.

• Enterprise level quality assurance is not free.

• Certification, compliance, indemnity is not free.

• Security patches and updates are not free.

• Commercial grade Linux is not free.

• Imposes top-down learning curve

• Long-term cost dominated by labor

• Back-port vs. roll-forward

• Upstream contribution vs. out-of-tree

• Product planning to co-utilize Kernels and QA resources

• Internal inter-product-team communication

• Centralization of bug reporting and fixing

• No re-invent, code-reuse

Page 14

Adoption: Long Term Cost


35pt

32pt

) :18pt


Company Overview

Conclusions





Page 15


35pt

32pt

) :18pt

• Linux is deployed in large number of Huawei products

• We plan to extend the usage of Linux as much as possible in all domains:

• Cloud Platform Network Gear Consumer Devices (mobile devices,

connected home and others)

• Huawei is investing a lot of engineering resources into the usage of Linux

• Different processor architectures

• Linux and Open Source (combined with commercial and own software) are

great tools

• Time-to-Market benefits (help building network gear faster)

• Quality, well known software

• Linux is a great commonality tool across products and R&D organizations (Architectural

benefit, common tools, experience sharing, R&D efficiency, R&D expenditures)

• Linux combined with multi-core hardware designs is crucial tool to keep

up with growing bandwidth demand (more on challenges slides)

• Huawei is investing into Linux Foundation

• Visibility and on openness image

• Participate into selected projects

• Advance the development of Linux

Page 16

How Linux and Open Source Can Help


35pt

32pt

) :18pt

Linux Enablers

Page 17

• Language barrier: Not applicable for source code • Communication via patches is Human Language insensitive

• Engineering staff enthusiasm • Engineering teams are relatively young and thus eager to

learn

• Top-level buy-in and support • Have seen helpful top-level changes in just 8 months

• Need a LOT more of that

• Existing Linux-based product portfolio

• General industry-wide Linux adoption


35pt

32pt

) :18pt

1) Open Source Software is great, leveraging Open Source Software is

key to delivering highly complex products to market on time

2) Availability of commercial, high quality, standards based software

components is great as well

3) Huawei is increasingly using open source software

Time-to-market

R&D efficiency

Managing R&D expenditures

4) Commercial software is used as well as own developed software,

naturally … we use all of these and benefit from them !

5) Huawei has good Open Source process in place

6) Contribution is very important. High quality contributions to open

communities will pay back very quickly and benefit the company and

create an openness perception of the company

7) Continuous education is going on at the company (on all levels) on

Open Source

Page 18

Open Source Conclusions


35pt

32pt

) :18pt


Company Overview

Conclusions





Page 19


35pt

32pt

) :18pt

• code sharing and re-use

• code-push-back vs. internal maintenance

• difference between IP and platform context

• you can change anything but maybe shouldn’t

Page 20

Continuous Education: Understanding


35pt

32pt

) :18pt

• Top-down vs. bottom-up culture clash

• Environment challenges

• Policy challenges

• Re-invention, Silos, Communication

• Proliferation to OSS values and principles

• Access to development infrastructure

• Interaction with external community

Page 21

Continuous Education: R&D Challenges


35pt

32pt

) :18pt

The Ecosystem Formed…

Page 22

The Result…!

HDFS

(Hadoop Distributed File System)

HBase

(Column DB)

MapReduce

(Job Scheduling/Execution System)

Zo

oke

ep

er

(C

oo

rdin

atio

n)

Pig

(Data

Flow)

Hive

(SQL) Sqoop

Avro

(S

eria

liza

tio

n)

*Non generic customization

We are contributing…!

Huawei Contributing Areas

Apache Software

Foundation

Custom

Features*

Information

Security

Product

Collaborative

Development

Version

Management

Strategy

Defect Fix and

Patching

Strategy

Review and

Quality for the

code

Huawei Development Team

Open Source

Centre

Software

Management


35pt

32pt

) :18pt

Component

Defects Reported

Patches

Committers Contributed Committed

HDFS & Core

95 74 64 1

Map Reduce

105 88 53 0

Hive 39 48 29 0

HBase 60 69 50 1

ZooKeeper 26 29 11 0

Total 325 308 207 2

Acknowledgments This work has been a community effort from the start, and represents

the work of many contributors. Both the architecture and

implementation were the collaborative effort of many. In particular,

this work would not have been possible without contributions

from Todd Lipcon, Eli Collins, Uma Maheswara Rao G, Bikas Saha,

Suresh Srinivas, Jitendra Nath Pandey, Hari Mankude, Brandon Li,

Sanjay Radia, Mingjie Lai, and Gregory Chanan.


35pt

32pt

) :18pt

Linux Challenges in R&D

Page 25

• Linux Development on Windows

• Windows VM via VDI to access remote hardware

• Standard Linux/OSS tools missing / not easily accessible

• Mail patches from Exchange client

• Pull source code local

• Ability to work with local hardware • ‘Observe’ hardware, logic analyzer, oscilloscope for hw debug

• NFS root export and access

• Local jtag / physical debug

• Transfer / Export GPL code

• IRC protocol

• Pastebin


35pt

32pt

) :18pt


Company Overview

Conclusions





Page 26


35pt

32pt

) :18pt

Page 28

OpenHuawei.org R&D Lab

•Geoff Levand

• Nikita Shulga

• Peter Popov

• Sven Dietrich

• Vlad Buzov

• Yuri Frolov


35pt

32pt

) :18pt

Page 29

OpenHuawei.org R&D Lab

• Web service accessible by everyone at Huawei • via http(s) using the corporate proxy

• complying with activity monitoring

• no direct connection to the internal Huawei networks

• OpenVPN

• Common Linux/OSS tools for internal access • Git / Gitweb

• Subversion, Mercurial, etc.

• Bugzilla

• Mail lists & Mail Server

• IRC server

• Project management tools like Redmine

• TFTP, Bootp, NFSroot, u-boot support

• Linux Kernel build and debug tools

• GCC tool chain

• Wiki collaboration


35pt

32pt

) :18pt

OpenHuawei.org

Page 30

External OSS Development Environment • Google Apps Domain

• OpenSuse 12.1 Linux Infrastructure (with systemd)

• Global Access via OpenVPN gateway

• Internal DNS (masq)

• Shared (team) development server Dell • 4 x 10 HT cores

• 128 G RAM

• RAID

• Test / Development Platforms • HP blades 2 socket, 6x HT cores

• 4 x Panda Board

• RackStation SAN via iSCSI

• 10 Gb backbone to SAN

• 1 Gb front end to ISP gateway


35pt

32pt

) :18pt

Page 31

OpenHuawei R&D Lab Goals

• Help recruit Linux engineers

• Enable productive OSS development in the US

• Ability to work with local hw

• Enable development on native Linux workstations

instead of Windows

• Host/mirror upstream OSS source code used

• u-boot, Linux, Virtualization, RT patches, etc.

• Host a build service (OBS or Yocto)

• Host all developer home directories as needed

• Access to OpenHuawei products by Huawei worldwide


35pt

32pt

) :18pt

Linux Adoption: General

Company Overview

Conclusions





Page 32


35pt

32pt

) :18pt

OpenHuawei Linux Projects

Page 33

Isolated Processing Environment (IPE)

• Telecom Carrier

• Packet Switching / Routing

• Control / Signaling and Transport Plane separation • Fast path processing in the transport plane

• SMP Linux overhead (syscalls, timer tick, irqs, softirqs, etc)

• Dynamic shifting of cores between control and transport

• Per-core power management

• Bare Metal Performance desired


35pt

32pt

) :18pt

Native Linux Data Plane / Control Plane Solution

Linux CPU

Isolation

Technologies Linux - SMP

Management Plane

Core 1 Core 2 Core 3 Core 4

Linux CPU

Isolation

Technologies

• Linux process isolation technologies

• Fast path network stack in user space

• User space device driver based solution™

Huawei

Network

Engine

Huawei

Network

Engine

Page 34


35pt

32pt

) :18pt


Company Overview

Conclusions





Page 35


35pt

32pt

) :18pt

Thank you www.huawei.com

RT Embedded Virtualization · HUAWEI TECHNOLOGIES CO., LTD. 35pt 32pt ) :18pt Page 29 OpenHuawei.org R&D Lab • Web service accessible by everyone at Huawei • via http(s) using

Documents