Eisuke Ito, Eiji Abe, Yoshiaki Kasahara Kyushu University - - PowerPoint PPT Presentation

eisuke ito eiji abe yoshiaki kasahara kyushu university
SMART_READER_LITE
LIVE PREVIEW

Eisuke Ito, Eiji Abe, Yoshiaki Kasahara Kyushu University - - PowerPoint PPT Presentation

APAN29, Sydney 2010 UPKI update from Japan Eisuke Ito, Eiji Abe, Yoshiaki Kasahara Kyushu University ito.eisuke.523@m.kyushuu.ac.jp Outline 1. Introduction 2. Shibboleth SSO 3. Problems of IdP 4. Analysis 5. Conclusion 1 1. Introduction 1.


slide-1
SLIDE 1

Eisuke Ito, Eiji Abe, Yoshiaki Kasahara Kyushu University ito.eisuke.523@m.kyushu‐u.ac.jp

UPKI update from Japan APAN29, Sydney 2010

slide-2
SLIDE 2

Outline

  • 1. Introduction
  • 2. Shibboleth SSO
  • 3. Problems of IdP
  • 4. Analysis
  • 5. Conclusion

1

slide-3
SLIDE 3
  • 1. Introduction

1. Introduction

  • 2. Shibboleth SSO

3. Problems of IdP

  • 4. Analysis

5. Conclusion

2

slide-4
SLIDE 4
  • 1. Introduction
  • Protected services in University

 E-Learning, e-Syllabi, Researcher activity DB, Student

portal, …

 E-Journal, Google Apps, Windows Live, …

  • Shibboleth SSO (Single Sign-on)

 Distributed SSO Middleware  Identity Providers (IdP), Service Providers (SP). (and

Discovery Service (DS))

  • Federation

 A trust relationship between Identity Providers (IdP)

and Service Providers (SP).

 NII of Japan deploys Shibboleth SSO Federation  Kyushu University joins this federation

3

slide-5
SLIDE 5

History

2005 2006 2007 2008 2009 2010 2011

LDAP

kitenet (Wireless LAN)

University ID University authN plaBorm

Joined UPK Shibboleth IdP

Kyushu Univ.

Password manager

ID IntegraJon UPKI SSO trial UPKI‐Fed Trail eduroam.jp Server Cert. service UPKI IniJaJve GRID

UPKI‐Fed Federa7on NII, Japan

Join UPKI‐Fed (JP Federa7on) 4

slide-6
SLIDE 6

In this presentation,

  • Show a case study of shibboleth IdP and

SP operation in Kyushu University

  • Report some problems of shibboleth IdP
  • peration.
  • Report results of two month operation.

5

slide-7
SLIDE 7
  • 2. Shibboleth SSO

1. Introduction

  • 2. Shibboleth SSO

3. Problems of IdP

  • 4. Analysis

5. Conclusion

6

slide-8
SLIDE 8
  • 2. Shibboleth SSO
  • Before Shibboleth

 Kyushu U

 Students 18,000  Staffs 7,000 (prof. 2500)

 Campus wide authentication system (since

2007.)

 IDM and LDAP server

 IDM: Identity management system (Meta Directory)  LDAP server is used as password authenticator.

7

slide-9
SLIDE 9

Password manager Active Directory

Enterprise system for officials

Matrix code DB Active Directory

EducaJon system for students login

IC Card (User ID) ID card

login

Other worker

Secure Matrix code authN Critical Service Critical Service ID/PW, or Matrix code LDAP ref Shibboleth IdP (SSO)

IDM

Shibboleth SSO Mail WebCT EZproxy kitenet (WiFi) refer refer ID/PW

System Overview

8

MS DS Refworks Shibboleth SSO E‐Journal

Federated SPs

MyLibrary

Personnel DB Student DB

slide-10
SLIDE 10

Dataflow of IDM

Staff list

LDAP

Staff

Daily IDM (Identity Management Sys.)

Personnel DB Staffs

ID Card

Student list Twice in a Year

Student DB Students

Learning System (PCs, Server, WBT) Account activation

ID Card

9

slide-11
SLIDE 11

Integrated Services

  • WebCT (e-Learning)
  • NetAcademy2 (English study)
  • kitenet (WiFi)
  • Campus licensed software
  • Space management system
  • Cute.Anyware (E-journal proxy by Library)
  • Webmail (Primary e-mail service)
  • Course registration and grade point management sys.,
  • Researcher activity DB

10

slide-12
SLIDE 12

SSO Policy

Internal service Out sourced service Usability

  • riented
  • Webmail
  • WebCT (e‐Learning)
  • Software download site

(licensed software)

  • University portal
  • E‐Journal services
  • RefWorks
  • Google Apps

Security

  • riented
  • Financial system
  • Grade point

management system

Shibbolize!

11

slide-13
SLIDE 13
  • 3. Problems of IdP

1. Introduction

  • 2. Shibboleth SSO

3. Problems of IdP

  • 4. Analysis

5. Conclusion

12

slide-14
SLIDE 14
  • 3. Problems of IdP

SP

LDAP

IdP

Attributes matching Attribute filtering

13

slide-15
SLIDE 15

Attributes matching

LDAP

IdP

Existing schema (attributes) eduperson schema

Add/Change schema

Solutions

Attributes Translation Schema Matching

attribute-resolver.xml

mismatch

OpenLDAP’s rewrite module

Open LDAP (rewrite)

14

slide-16
SLIDE 16

Attribute filtering

SP

IdP

MyLibrary EZproxy WebCT

Internal SPs SP SP OK Serves all attributes SP External SPs SP SP No Against privacy policy

15

slide-17
SLIDE 17

IdP SP

Solutions for attribute filtering problem

Internal SPs

  • 1. Write filter rules for each SP.

SP SP SP External SPs

rule rule rule rule

IdP SP SP

  • 2. Two IdPs

rule

IdP SP SP

rule

LDAP IdP Open LDAP (rewrite)

  • 3. Filtering script

rules

16

slide-18
SLIDE 18
  • 4. Analysis

1. Introduction

  • 2. Shibboleth SSO

3. Problems of IdP

  • 4. Analysis

5. Conclusion

17

slide-19
SLIDE 19
  • 4. Analysis

18

SJSDS (Solaris)

Shibboleth IdP Tomcat (CentOS) (VMware) (Windows XP)

OpenLDAP, CentOS.

LDAP IdP Open LDAP

(rewrite)

SP MyLibrary

EZproxy WebCT

Internal SPs SP SP Shibboleth IdP and SPs in Kyushu Universiy

slide-20
SLIDE 20

IdP SP

19

Kyushu University Library http://www.lib.kyushu‐u.ac.jp/

slide-21
SLIDE 21

Results

  • Two months operation

 Just serviced in at Dec. 1, 2009.

  • No serious trouble.
  • Some trivial matters.

 Some users bookmark the IdP site.

 404 Not Found: /idp/Authn/Password 7time(s)  He/She can’t access to the service which he/she wants.

20

slide-22
SLIDE 22

Statistics: Unique users

Students 1815 18000 Staffs 467

7000 (2500)

Total 2291 25000 at Jan.26,2010.

21

slide-23
SLIDE 23

Statistics: Rank‐Freqency

Top 200 users (10% users)

  • ccupy 41.2% access.

22

slide-24
SLIDE 24

Statistics: Daily access

23

slide-25
SLIDE 25

Statistics: Hourly access

Most user access at afternoo.

24

slide-26
SLIDE 26
  • 5. Conclusion

1. Introduction

  • 2. Shibboleth SSO

3. Problems of IdP

  • 4. Analysis

5. Conclusion

25

slide-27
SLIDE 27
  • 5. Conclusion
  • A case study of Shibboleth IdP in a

university

  • Two problems for IdP construction

 Attribute matching  Attribute filtering

  • Two months operation

 No serious trouble.  Got statistics

 No over load

26

slide-28
SLIDE 28

27

Thank you for your attention.