Evolution of Inter-Cloud Computing to Handle Big Data June 7, 2013 Tomonori Aoyama

Chair of GICTF Keio University

5th China Cloud Computing Conference 2013

Inter-cloud computing allows on-demand reassignment of cloud resources including compute, storage and network, and transfer of workload through interworking of multi-national and multi-operator cloud systems.

What is Inter-cloud ?

Private Cloud

General UserSmall Company

Public Cloud

Provider

Single Cloud2010 Hybrid Cloud 2011 Inter Cloud 2015

Big Company

Government Organization

Private Cloud

General UserSmall Company

Public Cloud

Provider

Big Company

Government Organization

Public Cloud

Public Cloud

Private Cloud

Private Cloud

Network

Evolution of Cloud Computing

From Single Cloud to Hybrid Cloud, which is a connection

between Public and Private Cloud, and then to global

Inter-Cloud Computing

T. AOYAMA 3

More than Zetta Byte Data are being produced. Enterprise Data, Video Data, Sensor Data,

Open Data Policy Personal Number Law (Japan) was approved.

Medical Welfare Cloud for EHR & PHR My hospital anywhere is targeted. (Japan)

I N TER CLOUD

Why ? Three Data require Inter-Cloud

Annual global IP traffic will pass the zettabyte threshold by the end of 2016,

Internet Video Will Drive Most Consumer Internet Traffic Through 2016

Sources: http://www.wired.com/magazine/2010/08/ff_webrip/ Cisco estimates based on CAIDA publications, Andrew Odlyzko

email

ftp P2P

Web

Video

2000 199

0

2010

54%

Internet video

(86%, the sum of all

forms of video (TV,

VoD, Internet, and

P2P) )

Sources:http://www.cisco.com/en/US/solutions/collateral/ns341/ns525/ns537/ns705/ns827/VNI_Hyperconnectivity_WP.html

T. Aoyama 5

Provider Y: Community cloud

Internet

Provider X

Avoid additional delay caused by the

increased load by means of

alternating data for access point to

other cloud provider

overload

Lease the required

quantity of

resources from the

community cloud

Service consumer

Network

Distribute the load on the own

servers to outside the own cloud

system

Numerou

s access

attempts Exchange data such as ID

information of consumer

Return the processing to the

original cloud system when load

distribution becomes

unnecessary

Service 1

Service 1

Service 2

Abrupt

increase in

the

number of

access

attempts

Use case 1: Guaranteeing performance against an abrupt increase of the load

DB

Cloud A

Cloud D

Cloud C Service 1

Disaster

Service 2

Service user

Continue to provide services by recovering each service of Cloud A by using a cloud system other than that of Cloud A

Select a cloud system to be used for service recovery through negotiation with other cloud systems

DBs are copied and synchronized in advance

Remotely activate the application for Service 2 (some quality requirements are satisfied on a best-effort basis)

Remotely activate the application for Service 1 (to guarantee the required quality)

DB

AP

MW

Service 1

Service 1

Service 2

Service 1/2

Service Y

Service X

Service 2

Cloud B

AP

MW

Use case 2 : Disaster Recovery

7 GICTF

8

GICTF: A forum for the Inter-cloud Technology

Applications

Network

Servers

Storage devices

Cloud system A

Standard inter-cloud interface

Cloud system B

Applications

Internet, NGN, SDN.

Application

Promote R&D on Inter-cloud technologies and the global standardization

through collaboration between academia, government and industry

Established on July 17th 2009

Member: 86 organizations ( NTT, KDDI, NEC, Hitachi, Fujitsu,

Toshiba, Microsoft, IBM, Oracle,

Cisco, VMware, IIJ, BIGLOBE, NICT,

NII, NRI, etc. ), 38 members

from Univ.

Medical

Network

Server

Storage

Application

Provisioning

Monito

r and C

ontro

l

Cloud System A

(A) Inter-cloud interworking architecture and standard interfaces

CDN

CDN e-Gov

Cloud System B

Provisioning M

onito

r and C

ontro

l

Legacy System

(C) Provisioning, monitoring and control across multiple clouds

Finance

Medical E-Gov

(B) SLA assurance for totality of server, storage and NW

Key Issues for Inter-Cloud Service Federation

Our target is an environment in which there are more than 10 distributed clouds, each

consisting of several hundreds of applications and several thousands of virtual servers.

Finance

10

Functional Structure of Inter-cloud systems

Intranet

Servers and storage

Monitoring and control

Network Intranet

Servers and storage

Monitoring

and control

Provisioning Service provision requirements

Monitoring information

Resource plan

Search for available resources Policy negotiation

Network configuration / optimization

Resource reservation request Flexible resource allocation,

reconfiguration

Provisioning

Inter-cloud monitoring and negotiation control

Dynamic reconfiguration of the server environment

Dynamic reconfiguration of the network environment

GICTF

The Worlds First Public International Standards Setting on Cloud Network

7 frameworks of public international standards are

almost completed

I. Cloud Computing Definition Vocabulary definition, etc.

II. Requirements Service requirements Use cases, etc.

III. Cloud Network Network structures Network requirements, etc.

IV. Cloud Computing Infrastructure Structures of infrastructure Infrastructure requirements, etc.

V. IP Resources Management Ability of resources management, etc.

VI. Inter-Cloud Method of inter-cloud Security management, etc.

VII. Cloud Computing DaaS Requirements and architecture of

Desktop as a Service

Major Deliverables

I. Use case and functional requirements

for Inter-Cloud Computing E/Aug 2010

II. Inter-Cloud interface specification on

protocols J/Dec 2011, E/Apr 2012

III. Inter-Cloud interface specification on

resources data model for network

control J/Dec 2011, E/Apr 2012

IV. Network and technical requirements in

support of Inter-Cloud J/Dec 2011

E/Apr 2012

Proposal based on

the results

http://www.gictf.jp/index_e.html 11

Framework of Inter-cloud Interface Specification

The interface between two cloud systems administered by different operators

Three layer modeling Lower layer protocols assumed as some XML message exchange,

e.g., REST or SOAP

Inter-cloud protocol: Information flows, message semantics with associated parameters specified

Data models for network resources specified

Data models for computing and storage referenced to other SDOs specifications

Inter-Cloud Protocol

Lower Layer Protocols (e.g., REST or SOAP)

Data Center Resource Data Model

(Computing and Storage)

Network Resource Data Model (Network)

Inte

r-C

lou

d IF

Cloud System

Data Center Resource

Network Resource

Cloud System

Data Center Resource

Network Resource

Inter-Cloud Protocol

PaaS API

SaaS API

IaaS API

Implementation

BSS/ OSS

Security

Open Cloud Manifesto

CSA

ENISA

DMTF

OASIS

ETSI

Inter-Cloud

Other Cloud

TM forum

ISO/IEC JTC1

ITU-T IEEE

DMTF

ITU-T

OCC

Taxonomy, Use cases, SLA, Requirements

NIST

NIST

Architecture ITU-T NIST

SDO coordination OMG IEEE

ITU-T

OCC

SIENA OGC CBA

KCSA CCF

GICTF

Japan Asia EU USA Global

GICTF

Cloud

OGF

SNIA

DMTF

ATIS CSF

OpenStack

Users/

Operators

Cloud Mng

API

ISO/IEC

JTC1

Cloud Standardization Status

SLG Special Liaison Group member

OGF OMG Cloud Interoperability

Roadmaps Session

ITU-T FG Cloud

CC Standard Study Group

15

NIST CC forum

OGF-

Europe

DMTF SNIA

OpenStack

Open Source

Community

DE-facto Standard

US Government

De-jure Standard

Cloud Business Google, Salesforce, Amazon, etc.

ISO/IEC JTC SC38

15

Collaboration among GICTF and various SDOs

T. AOYAMA

Inter-Cloud System R&D* (2009-2012)

Cloud Resource

Federation and

Reconfiguration

Cloud Resource

Provisioning

Real-time Sensor

Node

Dynamically

Reconfigurable

Network

*Funded by Ministry of Internal affairs and Communications of Japan

Physical World

WAN edge

Network

Cloud System A

Cloud System B

Server

Storage

AP AP AP Application

Wide Area Network

JGN-X Tohoku-1

JGN-X Tokyo-1

Base(1) Tokyo

Base(2) Tohoku

JGN-X Kyushu-4

Exclusive Line100M

JGN-X Kyushu-1

Inter-cloud technology evaluation will start in 2012 on the national-scale test bed where three cloud data centers in Tokyo , Tohoku, and Kyushu will be connected with each other via JGN-X.

Inter-cloud test bed planned in 2012

Base(3) Kyushu

Cloud C

Cloud A

Cloud B

Cloud Computing Testbed

New Generation Network Testbed JGN-X

R&D Testbed domestic companies and universities

Wireless Networking

Technologies Optical Networking

Technologies

Inter-Cloud

Technology

Advanced Application

Technology

companies &

universities

International

cooperation

IEEE Cloud Testbed

Testbed for Cloud Networking will be an essential part of Inter-Cloud technology developments

Coordinated developments through open test environments will be mutually profitable for both cloud computing and networking like New Generation Network

International cooperation with various organizations in the industry for the efficient development of Inter-Cloud standards

Promotion of cloud networking technologies and gaining the momentum

18

Virtualized Network

Necessity of Inter-Cloud Testbed

Cloud Reserch Institution

Cloud Reserch Institution

Emerging Areas of Study in Future Network Research

Network Virtualization (NV) Software Defined Network (SDN) Network Functions Virtualization (NFV) And more?

19

Introducing programmability into networking to flexibly and dynamically resolve constantly arising contemporary issues.

The University of Tokyo Confidential

Publication at SDN and OpenFlow World Congress in October 22-24, 2012

SDN (Software Defined Network) and NFV ( Network Functions Virtualization)

SDN Standardization

22

DPN: Deeply Programmable Network

Application Programmability

Control-Plane Programmability Interfaces Functions

Route Control Access Control Network Management

Data-Plane Programmability Interfaces Functions

Packet Data Processing Network Appliances (DPI, BRAS, EPC) In-Network processing (Cache, Transcode) Wide-Area generic processing

Handling New Protocols IPvN (N>6), New Layer2, CCN

The University of Tokyo Confidential

Control-Plane Elements

Network Applications

Interfaces

Data-Plane Elements

e.g, OpenFlow Controller

e.g, OpenFlow Switches

e.g, Southbound API

Interfaces

e.g, Northbound API

Applications

Control Plane

Data Plane

ONF OpenFlow

SDN

OpenDaylight

NFV

DPN

Deeply Programmable

Network (DPN)

Internet

Architecture of VNode Prototype

DC

R

P

R

P AGW

AGW Users

Users

Internet Programmer

R

P

Redirector (R) Programmer (P)

Slice Traffic

Slow Path

Fast Path

VNode Manager

SP

SP

SP

SP

SP

FP

FP

FP

FP

FP

Forwarding Engines

Data Plane

Tunnel Redirection

Redirector Manager Programmer Manager

Redirector

Access Gateway (AGW)

VNode

Domain Controller

23

Production Router (10Gbps)

VNode Architecture

User Opt-In

User Opt-In

Form he Skelton of Virtualized Networks

Programmability in Virtualized Networks

T. Aoyama

JGN-X Network Figure (from 2011/4)

USA

Thailand

Singapore

Hong Kong

40G 10G

DF

VN Virtual Node

OF Openflow

DCN DCN

Ex

Realizing multiple New Generation Network Planes on a virtual JGN-X network. Also made available is to interconnect with other Testbeds.

VLAN Testbed Network

VN

VN

VN

VN

OF

OF OF OF

OF OF

DCN

DCN DCN

V-node Plane

Openflow Plane

DCN Plane

Optical Testbed Wireless Testbed

1G Virtual Storage

StarBED

To Korea

Optical Testbed

Koganei-Otemachi OtemachiOtemachi

Bandwidth International

Japan- USA 10G Japan - Korea 10G Japan Hong Kong 10G Hong Kong - Singapore 2.4G Singapore - Thailand 622Mbps

Wireless Testbed

Fukuoka

Hiroshima

Okayama

Osaka

Nagoya

Kanazawa

Sendai

Tokyo NICT Koganei

40Gx2

40G 40G

10G 10G

10G

10G

VN

VN

VN

OF

OF

OF

OF

OF

OF DCN

DCN

DCN

Sapporo

10G

10Gx2 10Gx2

10G

1G

OF

JGN-X Planned POP

Kashima Hakusan Yokosuka Kobe Hokuriku

Keihannna Tsukuba Kyusyu Univ. Kyusyu Inst. Tech Tohoku Univ.

Tokyo Univ. Osaka Univ.

DCN

LA, USA

T. Aoyama 24

Social Cloud Infrastructure w/ SDN

Security and Data Portability in Virtualization & SDN env.

Integration of Inter Cloud and

Network Virtualization

Versatile Information Services

Total Handling and Control Mechanism

T. Aoyama 26

Virtual Network Resource Virtual Server Resource Virtual Storage Resource

Technical Issues on Network Virtualization for Intercloud Cooperation

Resource Abstraction Hiding details of resources Name resolving of resources

Resource Isolation

Isolation among private virtual networks QoS control to guarantee bandwidth or latency

Programmability

Enabling creation of network functions based on new ideas (cache/encryption)

Elasticity Providing network resources on-demand Cloud networking among clouds

Acknowledgement & Credits: The white paper of Network virtualization study group in 2010.

Summary

Inter-Cloud is a sequent trend toward future ICT systems GICTF objectives

Promoting the open interface between cloud systems Contributing the Inter-Cloud standardization based on

technical feasibility studies

GICTF deliverables and contributions to SDOs Use case and functional requirements for Inter-Cloud

Computing

Technical Requirements for Supporting the Inter-cloud Networking

Inter-Cloud Interface Specification Draft Cloud Resource Data Model Inter-Cloud Protocol

Collaboration with DMTF and SNIA Support for joint projects and international intercloud testbeds

Global Inter-Cloud Technology Forum

GICTF http://www.gictf.jp/