Category: "ETL/EAI/ESB"

MuleCon2007 Virtual Tax Services

Dan Cahoon and Christopher Ginn from H&R Block are presenting their experiences with Mule.

[As an aside, Dave mentioned that Ross is over in the corner working on his slides for later. Ross yelled out that "they're coming". And on another note, later tonight, there will be 9 ball, 8 ball and darts competition with the room against Dan, Christopher, Ross and Dave - winner gets a Mule jacket.]

The goals of virtual tax services is to move the tools to where the preparers do the work. H&R Block have over 13,000 locations in the USA, more than those burger joints. They spin up for four months around their own mini-Y2K... tax season. The concept is to move this out with a massive virtual service bus as a real-time, asynchronous communications platform for all of those offices.

One major goal was to demonstrate the open source technologies as a viable alternative to commercial solutions. Given the conservative nature of the company, this is an amazing vote of confidence that open source solutions are ready for the enterprise.

Risk mitigation is often more important than functions or cost. One consideration was that for all of these nodes [remote, temporary offices], connectivity is all over the map. Trying to accomplish the goals with a annual licensing from a large vendor just wouldn't make sense.

Christopher went over the VTS Architecture, showing Mule at the office level, and how it interacts with the main Mule VTS Service Bus. Many of the applications using the bus are open source, such as Alfresco for ECM including scanned documents from the offices. A great endorsement "Mule is, hands down, the best decision we ever made."

From the Q&A...

Development started in November 2006, and was rolled-out to sixty offices before the end of January, 2007.

SOAP and Rest are being used, with an emphasis on Rest. SOAP is mostly sticking around because of third-party support. [Legacy issues already. /sigh]

They don't use any business logic within Mule. It's infrastructure, it does what it's supposed to do very well, and they want to keep it simple.

We're blogging from MuleCon2007 throughout today and tomorrow.

MuleCon2007 Mashup

Eugene Ciurana, Director of Systems Infrastructure at Leapfrog, is presenting on Enterprise Application Mashups with Mule ESB.

Once upon a time, Eugene did industrial robots - we'll have to make sure that he and Clarise get to talk, as she still talks about her Singapore days, doing robots for Motorola.

Case Study 1: eCommerce Site

The objective for a large, international retailer, was to make an Architecture that would last for 5 years, through a 10 months skunkworks, that wouldn't force system upgrades on business units. The prime directive was to acquire not build, and to go with "best of breed". This prime directive was in recognition that no ONE vendor is best in every area. There was also a clear separation of concerns between platform infrastructure and applications, such as ERP, eCommerce, CMS, etc. [BI, where's BI in this plan?]. &#59;)

The result was a mixed environment of over 10 vendors, 3 hardware platforms and geographically distributed. How to integrate all of this? The answer is an ESB, and they looked at commercial and open source alternatives. From the commercial, one vendor looked good, but with an insane price; their JMS component was included in the final solution. On the open source, they looked at:

  • Jboss
  • OpenESB
  • Tuscany
  • ServiceMix
  • Mule

Jboss and OpenESB requires you to eat their whole enchilada [my words], and it wasn't digestible. ServiceMix and Tuscany are built to meet JBI, not solve real world problems. Eugene wanted to keep flexible, and not commit to Java as the solution for every problem. Mule was the solution. The resulting architecture has eCommerce Suite, Order Capture, Order Management, Web Service, Single Sing-on, CRM, CMS, Product Information, B2B/EDI and BPEL hanging off of the Mule Services Backbone, bringing about a successful reference implementation. This is being used to support a customer facing web structure that has 50 MILLION hits per day. The team only had 21 days to build the environment and integrate all vendors' products. Working with the Mule team, the first Mule-enabled components were the eCommerce Suite and CRM, which only took 7 days. Integrating the other 8 vendors took 20 minutes. More about this can be found from Eugene's site.

Case Study 2: Download Store

The objective was to create a download store, with a geographically dispersed team in 60 days. The work is still ongoing, but Eugene went into some of the details and showed the architectural diagram.

We're blogging from MuleCon2007 throughout today and tomorrow.

MuleCon2007 Introduction Day 2

Dave and Ross welcomed the well-over 100 attendees for MuleCon2007 Day 2. Let's just say that Dave should leave his day job for a stint in stand-up comedy. :D

Many of the ~25 MuleSource employees are here, and Dave introduced them. The staff is distributed among almost every time zone on Earth. They're looking for recommendation on team management software [They're living the TeleInterActive Lifestyle; ServiceCycle anyone?]

Here's some Mule stats:

  1. Over 600K downloads in 300+ enterprises.
  2. 58.3% of the users have build extensions to Mule.
  3. 67% want their current vendor to be more open.
  4. 47% feel the Jeda would win over the Ninja, only 8% thought superman would be the winner
  5. Approximately 400 pages of documentation on the website, but more is needed. Everyone has write access - sign up, and write away.
  6. Mule 1.4 & 2.0 - release dates at Ross' whims [Hey, it's open source], but they are planning more scheduling

The Mule team is always interested in understanding use cases - tell them what you're doing.

Mule is going international with a growing presence in Japan, UK and EMEA.


The MuleForge was developed to help manage the contributed code and extensions. The forge will also be a developer resource with components and use cases that don't belong in the main distribution.


MuleSource is looking for more partners to extend their reach. Contact Mike Lewis, the man in the pink shirt.

Thanks to Travis for the link from the MuleSource Blogs.

We're blogging from MuleCon2007 throughout today and tomorrow.

MuleCon2007 Training

The first half of the afternoon session, here at MuleCon2007, were simple labs inspecting and configuring the XML files for Mule, and seeing how they relate to the Java files. My MacBookPro can now ask your name, and then greet you by name, as a web service. :D

Peter is now going through modeling of applications, using order processing as an example. The applications to be integrated include a Sales tool, in Java, the Warehouse [physical not data] in COBOL, a Logistics application, Accounts, and the Assembly process. Since the Assembly process has no application in place, we'll create a service to track this process and facilitate interaction with the COBOL warehouse application via flat files, and with Logistics via JMS.

Think about the power inherent in that ability to interact with legacy applications.

There's a lot more in the online documentation from the Mulesource wiki.

And now it's back to more labs.

Mulesource and Ricston have put together a great day. There are about 100 people here from all over the world. This is a true testament to the interest and reach that open source projects and companies are now generating.

We'll be blogging from MuleCon2007 throughout today and tomorrow.

MuleCon2007 Routers

Types of Routers

  • Inbound
  • Outbound
  • Response
  • Filters
  • catchAllStrategies

One can stack routers & transformers

Peter gave examples of the ease of creating various special purpose routers, such as a wiretap, exception based or chaining router with the familiar XML configuration, showing the ease, power and flexibility of the Mule router implementation. Also take a look at the EIP book.

We'll be blogging from MuleCon2007 throughout today and tomorrow.

MuleCon2007 Recovery Strategies

Exception Strategies

Component Exception Strategies

Connector Exception Strategies

Essentially route/log

Fatal exceptions such as a message queue breaking also can be handled

Reconnection Strategies

Configure how a connector behaves when it connection fails

  • retry # & frequency
  • other basic reconnect strategies are included
  • create your own

We'll be blogging from MuleCon2007 throughout today and tomorrow.

MuleCon2007 Messaging Style

Asynchronous Inbound & Outbound

Asynchronous Inbound & Synchronous Outbound

Synchronous Inbound only

Synchronous Inbound & Outbound

Synchronous Inbound & Outbound with transaction

Request-Response allows a UMO to make a request and await a specific response

See diagrams for each of above from the to-be-published presentation [URI to come]

In mule, one can specify how many instances of a UMO can be invoked and configure various message threads to create a complex flow

Mule can automatically match incoming & outgoing messages or fkn support event calls throug its callable interface.

One can control synchronicity through the endpoint property

We'll be blogging from MuleCon2007 throughout today and tomorrow.

MuleCon2007 Theory

Most configuration in the XML file can be done either by setting a property or using a URI

Configuring JMS, embedded ActiveMQ

Using Queues & Topics

Selector Configuration Options

Durable subscribers can be configured

SOAP Core Transport

Hosting a web service with Mule can be done using AXIS orXfire - Example of simple HelloWorld POJO exposed as a web service - configure lnbound router & endpoint with transport

Accessing a web service with Mule can also be done using either AXIS orXfire; the same approach is used whether local or remote for AXIS but WDSL is used for remote access via Xfire

Over 25 Transports includeded with Mule; configuring one is much like any other. Mule also provides a framework to easily add other componts

With Xfire one can use the client, but more common is to see a series of chain routers.

The Big Plcture - see diagram - Endpoint [message receiver] -> Inbound Router -> Interceptor -> Inbound Transformer -> Service Invocation -> Interceptor -> Outbound Router -> Outbound Transformer -> Endpoint [message dispatcher]

Interceptors - see diagram from the presentation [link TBS]

We'll be blogging from MuleCon2007 throughout today and tomorrow.

MuleCon2007 Overview

Peter Delia, co-founder with Ross Mason [CTO of Mulesource], of Ricston, based in Malta, is providing today's training, starting with an overview of Mule, Mulesource and Ricston.

SOA Topologies; Mule can be configured to implement any topology

  • ESB - see Mule project front page
  • ESN
  • Peer Network
  • Client Server/Hub & Spolke
  • Pipeline

Mule Architecture: Each Mule instance connects to many Applications and/or Service Components & other Mule Instances; each Mule Instance has a Mule.Manager & Service Containers; more detail on wiki; Mule supports a wide variety of transport providers.

Core Concepts - see diagram on wiki

  • Service Components - can be simple POJO
  • Endpoints [message receiver or dispatcher, connector, transformer] & Transport Providers [synchronous & asynchronous endpoints possible]
  • Inbound Routers
  • Outbound Routers
  • Transformers
  • Service Manager

Question - running threads in Mule, running Mule on slower, but multi-core machine - it depends on specific needs and processes, as to whether or not it is advisable, but it is possible.

Configuring Mule

  • extensible
  • XML file
  • DTD
  • root is UMOManager [Universal Message Object = Service Component]

XML Configuration File Structure

  • Mule Environment
  • Connectors - used to send & receive data over a particular channel: name, className, 〈ref〉 - it is possible to configure connectors on endpoints
  • Transformers: name, className, returnClass, property
  • Endpoint - used to control how events are sent & received, best configured on routers; transformers are configured on endpoints; Global endpoints, Endpoint Identifier
  • Routers - inbound Routers, outbound Routers, response Routers, filters, catchAllStrategies
  • Service Components: mule-descriptor element
  • Mule Components

We'll be blogging from MuleCon2007 throughout today and tomorrow.

Talend - new open source ETL with a new data management vision

We would like to thank Charly Clairmont for visiting our Open Source Business Intelligence Squidoo Lens and writing us the following note...

"Talend is a new open source ETL with a new data management vision.
Talend's solutions collect, process and consolidate data residing in organization's various, heterogeneous systems and injects them into a centralized data warehouse (datamart)...

Talend provide 4 solutions :
* developer tools : to create process
* administrator : to manage distributed process on a grid architecture
* launcher tools : to launch process
* PAM : Process Activities Monitor

I just want to inform you about the avability of Talend a new open source ETL.

The ETL language is PERL, and JAVA. But Perl provide many more connectors than other java libraries.

What is very "sexy" is the client tool for developer which is build on Eclipse RCP. Also, because of etl experience of the leaders of the project, Fabrice Bonan and Bertrand Diard. The developer tool has a business and technical aspects...

You can find more information on :

Best regards.

end quotation

We have updated our Open Source Business Intelligence Squidoo Lens Links to OSS ETL Tools to reflect this. We look forward to checking out Talend.

Please feel free to contact us , if any of you know of any other Open Source BI Tools that is currently not in our lens or have any updates to the projects listed, any OSS events or any other feedback you would wish to give. &#59;D

October 2019
Mon Tue Wed Thu Fri Sat Sun
  1 2 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29 30 31      
 << <   > >>

At the beginning, The Open Source Solutions Blog was a companion to the Open Source Solutions for Business Intelligence Research Project, and book. But back in 2005, we couldn't find a publisher. As Apache Hadoop and its family of open source projects proliferated, and in many ways, took over the OSS data management and analytics world, our interests became more focused on streaming data management and analytics for IoT, the architecture for people, processes and technology required to bring value from the IoT through Sensor Analytics Ecosystems, and the maturity model organizations will need to follow to achieve SAEIoT success. OSS is very important in this world too, for DMA, API and community development.

37.652951177164 -122.490877706959


  XML Feeds