Enterprise Sakai Creating a robust Sakai infrastructure

donkeycheerfulInternet και Εφαρμογές Web

7 Αυγ 2012 (πριν από 5 χρόνια και 3 μήνες)

393 εμφανίσεις

Enterprise Sakai: Creating a
Robust Sakai Infrastructure


Mr. Josh Baron,

Director, Academic Technology, Marist College

Mr. Chris Coppola,

President, rSmart

Mr. Howard Baker, Project Manager, IBM

Mr. Austin Schilling,

IT Executive Consultant, IBM

MARIST COLLEGE


We are
NOT

a large research university!


Founded 1929


small complex
liberal

arts college


Approximately
5700 FTE

student
population


200 full
-
time faculty
, 500 part
-
time


Long history of
incorporating technology

into the teaching and learning process

Why do we need an

enterprise solution?

Answer: We’ve learned our lesson
from the past…

Learning from the past…


In 1996 the
digital horizon

was foggy…


We have a better view today…
watch out
!


Use and production of digital media will
grow exponentially

over next decade:


Electronic portfolios



Rich media learning objects



Social Podcasting


Next Up:
Mobile authoring tools
?

Colleges and
Universities

A National Learning Grid

K
-
12 District

State and
Local
Consortia

Museums,
science
centers, etc.

Certification
Exams

Online
Courses

“Raw” Video
Content

Student
content

Schema for Provisioning the Sharing of Resources/Tools

Sakai Repository


Marist/IBM Joint Study


Well Established Relationship, First Project


1988


Living Lab for New Technologies


Research projects, IBM Academic Initiative support,
Marist grant initiatives


IBM program manager provides overall project
management


IBM researchers, developers, business execs, and
Marist faculty, IT staff, and students collaborate on
numerous projects


Marist students hired as IBM interns


Opportunities for full
-
time positions


Marist Sakai Goals


How do we create a scalable, enterprise
version of Sakai capable of handling future
xxx needs?


Port Sakai to IBM WebSphere (WAS) and DB2

Global Campus

Students

Educators

Administrators

Customized Portals

App 1

App 1

App 2

App 2

App 3

App 3

C
ollaborative
L
earning
E
nvironment

Sakai Tools &

Development

Environment

Distributed Environment

Core Services
: content, user, security, site

Core Infrastructure

Sakai Vision

WAS Porting Issues


Classloaders


Tomcat decided to go against specification


Looks in local classloader first unless resource
contains certain package keywords (javax,
org.xml.sax, etc)


If resource not found, request is delegated to parent
classloader


WebSphere is the opposite, but allows the
administrator to adjust the classloading policy



WAS Porting Issues


Deployment Descriptors


Tomcat is much more lenient with invalid
deployment descriptors


If a web.xml file does not match the servlet
specification it overlooks the error and allows the
web application to function normally


WebSphere is much more strict


Will not allow an application to be installed unless
there are no errors in the web.xml file

WAS Porting Issues


Filters


The Servlet Specification states that any filter
that maps to a url
-
pattern should be applied
before any filter that maps to a servlet name.


Some Sakai components do not follow this spec


Tomcat applies filters in the order in which the Sakai
Application is expecting


WebSphere applies the filters according to
specification

WAS Porting Issues


Need to create a single code base



Easy to install


regardless of app server


Alternative build specified by global settings
variable. One code base has necessary source
for all versions.


XML specifies build


Source files


Destination of compiled source



DB2 Porting Issues


Work Involves:


Creating db2 versions of all sql/ddl scripts


Reworking issues from hibernate and legacy jdbc
services


Common Hibernate Gotchas


Long names for indexes


Binary blob types


Solution: Create hibernate dialect. This avoids
changes to java source or hbm files.

DB2 Porting Issues


Legacy JDBC Gotchas


No vendor specific abstraction in Sakai for
dealing with vendor specific issues


Solution: rSmart is working with the
community to move to vendor specific
helper classes that can be set at runtime to
make code more manageable and less error
prone.

Marist iLearn
x

Near Term Status


Port to WAS and DB2 near completion


June and July


Migrate courses and test


August


Create production infrastructure


Blade server based


September


Move into production in Marist’s z/OS Certificate
program


NOTE:
iLearn

= Innovative Learning Environment

and Resource Nework

Marist


rSmart


IBM
Contribution


QA environment running DB2 and WAS will
be live in September with several “flavors” of
Sakai.


Hosted at Marist College


Marist will also be contributing several QA
resources for testing activities


COME JOIN US!

Business Value

Ability to Dynamically Respond

Reduce costs

Increased
utilization

Increased
flexibility

Better IT


Public Service
Goals alignment


Future Directions
-

Dynamic Infrastructure to deploy,
maintain, and support a Utility based model delivering Education
Services to Academic Communities

Simplify:
Consolidate IT assets

Virtualize:

Create logical asset pools

Provision:
Automate
Capacity & Workload Mgt

Orchestrate:

Achieve Policy
-
based computing

IBM Dynamic Infrastructure










Tivoli Intelligent Orchestrator

IBM DI On Demand Workplace


Demo
Topology

De
-
/Provision

App/Portal Server

Cluster

Usage

Utility Business

Services

Reporting

Objective Analyser

ODW OA & DAE

Tivoli

Directory Server

(LDAP)

WES

(ND)

User / Robot

UBS

Filter

HTTP

Proxy

Metric Service

Server Pool

Update ND

Configuration

IBM Tivoli Provisioning

Manager

Workflow

Data

Center

Model

get/set

Plugin

DCM

Plugin

...

Plugin

Workflow

Workflow

A

A

A

Subscribe:

Premium/Normal

Service

Evaluate Usage Context

WebSphere SOA Component Clusters

Solutions/Apps

WPS

Sakai CLE

Web Services

J2EE Apps

rSmart OSPI

SeHS

J2EE Portal

ePortfolio


DB2

Domino


SeHS/WPS

Product


Stack

Metering

Aggregation/

Correlation

Two kinds of users:

a) “premium”: subscribed to get classified requests based on SLOs

b) “standard”: no SLOs

SPP & IDI Integration @ Marist


Approach


Stage 1 (SPP PoC)


Start with SPP for base provisioning


Integration of Slackware OS provisioning


TPM installation and setup


Stage 2 (IDI for Education Landscape Mgmt PoC)


Implementation and deployment of IDI


Leveraging SPP base provisioning


Implementation of Marist Education services Application Landscape
Managment (Sakai)


Stage 3


Converged SPP and IDI solution for Marist Education Services
(Sakai) Landscape Management


Improved error recovery and compensation

Q & A

Technical Contacts