Category: "Computers and Internet"

Users on Mule

The afternoon sessions are split into a Developer Track and a User Track. I decided to attend the User Track. The first speaker was Shigemoto Fujikura of OGIS International speaking about the S2Mule project that they implemented. As part of an all encompassing project affecting dozens of system, the first target was an HRM system, with the first application being a mainframe skills management system that was being migrated to an ERP system. Microsoft Excel was used as a client by the executive staff against the old system, and they desired to continue to do so under the new system. By using SOA, they were able to define the system boundaries through the system interfaces, and were able to allow MS Excel to access the SOA network [SOAP] using VBA. They began the project with the 2005 open source Mule ESB project, as it allowed them to start small and execute trials to test out decisions quickly.

Searsar2 is a widely popular open source framework in Japan making Java dependency-injection simple with aspect oriented programming. The S2Mule project brings Mule into Seasar2.

Hiroshi Wada of the University of Massachusetts has been working with OGIS International to use modeling techniques to define non-functional properties in SOA separately from functional requirements to allow the reuse of services and connectors in different contexts. The Mule ESB has provided the test platform for their use of UML, business process modeling notation [BPMN], feature models and aspect oriented programming.

Partner and Customers at MuleCon2008

The rest of this first morning of MuleCon2008 seems to be devoted to partner and customer use cases. The afternoon will be breakout sessions separated into a user track and a developer track. This is in stark contrast to last year's MuleCon, where the first day was training on Mule, while the second day was filled with alternating sessions devoted to core developers talking about the roadmap and users telling their stories. This change, coupled with the higher attendance indicates to me the maturation of Mule and MuleSource.

Jahan Moreh, VP Technology of U1 Technologies spoke about high performance messaging using Mule. He started off giving talking about messaging standards such as Java Messaging Service (JMS) and Advanced message Queuing Protocol (AMQP). He gave a very good overview, almost a tutorial, on messaging in general and the trade-offs involved in high performance messaging.

Eugene Ciurana, Director of Systems Infrastructure at Leapfrog Enterprises, Inc. spoke on "Son of SOA: Resource Oriented Computing and Event Driven Architectures" PDF download. Eugene provided an entertaining look at the limitations of SOA and the challenges and rewards of implementing SOA at Leapfrog, from the perspective of Charlie the Farting Dog. Well, not quite, this was an example of the upcoming Tag from Leapfrog reading a book. Leapfrog is very driven to implement new systems in very short timeframes using best of breed components. Of most interest to me is the shift towards consuming resources, blurring the distinction between services and data sources. All components of a system are viewed as resources to be consumed synchronously or asynchronously. there is no distinction between data, objects or services and there is no dependency on a programming language or framework. Programs map logical and physical locations through identifiers in traditional computing models. Resource Oriented Computing defines resources through verbs and logical identifiers, much like REST, but not exactly. Eugene went into a good comparison of how Java, REST and ROC handle Resource, Identify, Resolve, Compute, Immutability. Eugene selected Mule ESB as his ROC Backbone (workflow, transactions, transformations, logging and routing) and for the Resource Container. Extending the web protocols to go beyond the four REST verbs is easy within Mule. The initial ROC implementation took about 60 days, with developers just writing POJOs and everything deployed as a single JAR file. The ESB backbone is actually as many identical implementations of Mule as he needs, thus he can scale the backbone horizontally as much as needed. To summarize Eugene's conclusion, complex systems are easier to code and maintain if implemented as small blocks which can be mapped as resources that can be consumed in a stateless fashion reducing implementation costs by 70% and maintenance costs by 30-40%.

Roy dela Paz of Biogen Idec [Nasdaq: BIIB] provided a "Customer Case Study: Move over Proprietary, Here's Come Mule". The BEA Weblogic Integration Platform v8.1 has been in use at BIIB for internal and external data integration since 2004, but the learning curve has been steep [very abstracted, proprietary, non-standard development environment], it's been unstable [Nagios scripts have helped here], and it hasn't scaled well. Mule came back into play through a "water cooler conversation" and investigation showed that Mule provided the required capabilities in a lightweight, stable, scalable package using standard XML and POJO development skills. Roy compared the complexity of Weblogic vs Mule as that of the Sideways Bicycle vs a normal single-speed bike.

MuleCon2008 Intro and Year in Review

Mahau, the Director of Marketing gave the introduction. There are more than double the number attendees this year as last. All members of the core development team are here.

Dave Rosenberg provided the year in review.

  • Four babies, all girls, have been born to the Mule "family" in the past year
  • MuleSource launched in Japan
  • Mule Galaxy for governance was released
  • Mule2.0 and Mule RESTPak coming soon

Dave began discussing that building good products is important, but building the right products is equally important. Customers have been successful in SOA when the use an ESB as a foundation. Governance [Mule Galaxy] is the next important step. Mule Saturn for data monitoring and MuleHQ for system monitoring & management round out the Mule offerings.

The MuleForge has a lot of new material from developers in 118 countries.

From the survey that MuleSource recently conducted with their users, while up-front cost is a consideration, the most important value from open source software is that the users get a feeling of control by having access the code.

When Dave talks about the growing ecosystem, it now includes over 40 partner companies.

Ross Mason, creator of Mule and CTO & co-founder of MuleSource provided the Mule Product Updates and Roadmap session.

A year ago, there was the mule project being transitioned from an open source project to an enterprise product, including the announcement of MuleHQ.

This year, the open source offerings are the Mule ESB, Mule Galaxy, Mule IDE and the MuleForge.

The Mule community edition will have major releases this quarter and another planned for the last quarter of this year and Ross presented the 2008 timeline.

Mule 2.0 has had a lot of architectural changes under the cover. More visible to the users will be schema-based configuration. There will be expression evaluation support [Xpath, XQuery, Groovy, Context information for routing, transformations, etc].

Mule Galaxy is a registry and runtime governance that is deeply integrated with the Mule ESB.

Mule IDE is Eclipse based.

Mule Enterprise 1.5 goes through an extensive QA, testing and platform certifications processes. It also includes MuleHQ that allows things like profiling, and Mule Saturn (beta) that allows visualization of the flow through Mule.

The community edition will have more frequent releases but will be potentially less stable.

Mule Enterprise 2.x will add some premium connectors, such as for Websphere MQ. Migration tools will be added to MuleHQ and Mule Saturn. The migration tools include even easier migration from Mule 1.x. One major change in 2.x is the addition of SOA Runtime, hot deployment of components. When one has a service running in Mule, and one changes the message format for a long running transaction, hot deployment allows a way to gracefully change over from the older to the newer format, without shutting down.

Mule Galaxy Enterprise provides for Clustering, Microsoft Office Indexes, PDF indexes, Workflow, Replication/federation, Premium (task based) documentation and QA.

Ross provided a few use cases.

The Mule IDE (promised to be fully functional and stable by the end of the year) allows one to create, debug and deploy new instances of Mule, while using Mule Galaxy to share artifacts and design time policies. While developing in the Mule IDE, one can extend the functions of the instance by getting components from the MuleForge.

Within Mule Galaxy, is something called Mule netboot, which is essentially a bootstrap node for Mule. Mule HQ can discover these nodes, which can be configured to download its configuration from a Mule Galaxy URL. Mule netboot then starts up Mule ESB. This is much simpler than the current Mule patch manager method of deployment.

Monitoring Mule with MuleHQ provides triggers and alerts for things like error and load conditions. Mule Saturn provide more of a business view, while Mule Galaxy provides the runtime governance as to who/what is using the service and how.

Ross is now providing a tour of MuleForge. One strong aspect of MuleForge is that it allows folk to create a connector, for example, and then release & manage it through the MuleForge. There are also proposals for projects, so that one can seek help in creating, or at least in determining interest in, certain connectors or extensions for Mule.

There is a certification process for projects

  1. Not Certified
  2. Community Certified
  3. Partner Certified
  4. MuleSource Certified - MuleSource will support that connector directly as part of a paid subsciption

MuleSource is now doing targeted distributions from MuleForge, for example, a REST pack that allows for ATOM publication that provides not just a code bundle, but also instructions, descriptions and definitions on using the pack.

Conclusion

As noted in our pre-OSTT thoughts, and in some of our OSBC notes, that over there have been some holes in the open source offerings, most notably around governance. The release of Mule Galaxy will be filling this hole. From last year's MuleCon I had noticed that there was a dichotomy among the users after the training, some who embraced the XML editing for configuring Mule and others who missed their wizards. The schema-based configuration and the seamless integration of the Mule IDE should help to satisfy both camps.

Dave came back up to garner suggestions for the ending campground. The audience suggested:

  1. .Net integration
  2. EDI transformations and using the MuleForge
  3. Dynamic transformations
  4. performance and test authoring for the full life-cycle
  5. migration from 1.x to 2.x
  6. using the IDE
  7. Internationalization
  8. BPEL
  9. use cases for Mule Galaxy including non-Jackrabbit, JBPM & Alfresco
  10. clustering and high availability
  11. Integrating various ESB/SOA/integration product lines and can Mule be used to bring a "one of everything" shop together

OSBC 2008 The Hidden Session

There's a blank space in the agenda for this time slot. A panel has been brought together to discuss if open source companies can survive the M&A. The panel includes Harold Goldberg, CEO, Zend Technologies, Ben Sabrin, VP, Sales & Business Development, Appcelerator, Zack Urlocker, EVP of Products, MySQL, and Rex Wang, VP, Product Marketing, Oracle.

The community, the employees, the product - everything could change.

Zend and MySQL are strong partners, as over 50% of development on top of MySQL is in PHP. Sun acquiring MySQL initially was upsetting, but more from the standpoint that the world needs strong, independent open source companies. The two companies are dedicated to supporting the overlapping communities through the transition and to maintain the direction of MySQL. Sun and MySQL, whether still being in the honeymoon period or indicative of long term cultures, are having open, supportive conversations.

Red Hat and JBoss had no such conversations, and the cultural shift was very negative, though the results may be positive.

Harold brought out that there are three types of M&A: acquire a business, consolidate markets, purchase a piece of technology. Open Source doesn't make this different.

Sleepycat was acquired by Oracle two years ago, and it was announced at OSBC 2006, and thus has been the longest in the transition. Sleepycat personnel were integrated with comparable groups within Oracle, e.g. Sleepycat engineers joined database engineering, marketing joined marketing, sales formed with sales folk from other recent Oracle acquisitions, a new sales group focused on embedded databases and small devices. Though this integration and dispersal happened over many months, and some processes, such as the release cycle are still done the Sleepycat way and may never transition to the Oracle way.

There are gradations of M&A transition & integration policies, ranging from complete absorption into the new culture, to maintaining the acquired company intact while the acquiring company learns the new business. Most acquisitions fail. It's too early to tell how these acquisitions will run, but a lot depends on the M&A experience and competency of the acquiring company.

OSBC 2008 Server Out Network In

I haven't been doing any "live blogging" this OSBC. Yesterday was very hectic. I did some interviews, attended some sessions, and had many great conversations. Today was less hectic, and I even got to pull out the MicroTrak to record a podcast with Brian Reale of ProcessMaker.

I actually got to this panel session early enough to grab a piece of table and set-up my laptop. This panel is "Open Source: Out of the Server, Into the Network" moderated by Larry Augustin, Managing Director, Augustin Ventures and comprised of Artur Bergman, Director of Engineering, Wikia, Fabrizio Capobianco, CEO, Funambol, Kelly Herrell, CEO, Vyatta, and Mark Spencer, Founder and CTO, Digium.

Much of the conversation in the first part of the panel centered around Cisco as a monopoly, pricing of "enterprise class" at three or four orders of magnitude over similar items from the the local electronics store, and adapting Linux for network use. What I heard was the commoditization of the network. The same discussion I was having with the CEO of a communications VAR 10 years ago. At that time, the industry was going through a sea change where the old way was to have system engineering as a free, value-added service as part of the sales process for large, multi-million dollar PBX, EDGE Router, telecomms & networking deals. The new way hadn't been decided yet, but the margins were being forced down to under 15% - commoditization. Today this panel was talking about COTS hardware running specialized open source software to replace network gear and PBXs - commoditization.

But where's the value? Fabrizio hit the nail on the head: community, the power of community and using open source to build a community. Why is there value in community? I think we can sum this up with customization, localization, and common needs.

Larry summarized it as a vertical market coming apart horizontally.

Great session all the way around.

Adaptive Planning at OSBC2008

I met with Bill Soward, President & CEO, and Greg Schneider, Vice President Marketing of Adaptive Planning. Adaptive Planning provides packaged solutions for business performance management (BPM), but I think that the best description of Adaptive Planning is that they provide focused BI tools, both on-demand as SaaS and on-premise as open source software, which is licensed under GPLv2.

Since its founding in 2003, Adaptive Planning has tripled its customer base year over year, and has grown to over 220 customers, with 90% of these being customers of the on-demand SaaS. However, approximately half of the largest customers use the open source on-premise version. Most of these customers are companies with 100 to 2500 employees. In toto, they've taken US$29 million in VC funding.

The open source version is available from SourceForge and SugarForge, and has been downloaded over 65,000 times in 85 countries. They have data that suggests that a significant percentage of these downloads have been by finance folk, not IT. These fits well with our experience that end-users are starving for simple BI tools that simply aren't being provided by IT in many cases. I told the "lake of data" story, and I'll repeat it again. Several years ago, we were giving our benefits of BI pitch to a VP of Marketing at a business unit of a Fortune 100 company. She stopped us and brought in one of her marketeers. This very non-technical fellow had recreated data warehousing using over 100 linked MS Access database applications that he called the "Lake of Data". It was amazing. It worked, and it filled their needs. This story pre-dates any significant open source options for BI, but it is exactly this type of person who might download Adaptive Planning Express.

Their community model is a bit different and I believe that it is very worthwhile. They build their community around partners with specific domain expertise. This expertise may be within a vertical niche such as health or the public sector, or specific geography. One example of this is the Adaptive Planning partner in Japan worked with their customer AKTIO to implement localized version of Adaptive Planning's open source budgeting, planning and reporting tools for 800 seats, on-premise at AKTIO. There are currently 65 partners working on extending Adaptive Planning in different ways.

Adaptive Planning is also looking to partner with other open source companies, particularly in the area of data integration. When Adaptive Planning added reporting tools, they originally looked at using one of the open source tools. At the time however, the existing open source reporting tools were very much for developers, and not for end users. Adaptive Planning is very much about ease of use for the end users, so they decided to build their own dashboarding and reporting set. However, with the maturation of open source offerings over the past two years, Adaptive Planning will likely not build their own data integration tools, but will partner with an existing open source solution.

In many ways, the chief competitor for Adaptive Planning is MS Excel. That is, the end users choose the status quo of slogging their way through Excel workbooks and pivot tables. Bill emphasized that the Adaptive Planning strength is allowing user collaboration on modeling, assumptions and background information in a secure, controlled way.

SourceSense and Microsoft Apache POI

I'm having lunch with Gianugo Rabellino, who is very excited about the new partnership between his company SourceSense, the leading open source integrator in Europe, with the little known &#59;) software vendor from Redmond Washington in the USA. The partnership centers on a new top-level Apache project, Apache POI, which is the open source file format reader and writer to create, edit and read Microsoft Office formats used in Excel, Word, PowerPoint and Visio, and supports Open XML.

Microsoft provides the detail in a press release, Microsoft and Sourcesense Partner to Contribute to Open Source, Apache POI to Support Ecma Office Open XML File Formats. Congratulations to Gianugo and Sourcesense.

WSO2 Interview at OSBC 2008

I met with Paul Fremantle, Co-Founder & CTO of WSO2 at OSBC 2008. Paul and WSO2 CEO Sanjiva Weerawarana both come from IBM Web Services architecture and founded WSO2 to provide an IBM level of quality of service in a simpler implementation. I was actually introduced to Paul and Sanjiva about an half-hour before our scheduled interview by our mutual friend Gianugo Rabellino. Here's some of the information that Paul provided about WSO2.

WS* for security & messaging is a major focus for WSO2, and WSO2 is filling some of the gaps we mentioned in our "suggested topics" for OSTT, especially in the area of exposing data through WSO2 Data Services and Governance in the WSO2 Registry:

  • RESTful not UDDI
  • ATOMpub
  • Social Governance: Tags, Comments, Ratings, subscribe via any feed reader to tags, versions, etc.

Independent security through WSO2 User Manager or standard APIs and applications

Fifty percent of the WSO2 developers times is spent in Apache projects, and the company is 95% engineering. This shows one of the strengths of Open Source in that the company is not heavy with GS&A costs.

As anyone reading this blogs knows, we feel that Community is the major strength of any Open Source company. The WSO2 Community provides requirements & bug fixes, primarily. One recent example of the value of the WSO2 community is where Hessian provided requirements & testing.

WSO2 is very much a platform company, much like Spring. WSO2 worked with Spring to have Apache Axis2 work as a component within Spring using Axiom (AXis Object Model).

  1. Spring WS requires Contract First
  2. The WSO2 framework for Spring allows POJO Code First

The Axis2 beta was published six weeks ago and a GRAILS contributor has already contributed a plugin to the community that is dependent upon the WSO2 web services framework for Spring.

Business models for Open Source companies are still being developed. WSO2 achieves monetization through subscription, training, developer support [email & IM] and product support. All WSO2 products fall under Apache Licenses.

In addition, WSO2 provides Open Source Development: customers pay to get customizations into the main tree of the code base of Apache projects that are important to them. This is a great way to provide support to your favorite open source project.

We ran out of time at this point. In conclusion, we feel that WSO2 is filling some of the gaps in open source products for SOA, as well as having some unique approaches in extending their business model.

OSBC 2008 Starting Up

Clarise and I have registered and we're just getting coffee. :p We've already run into a few folk from the Open Source Think Tank. Matt Asay's keynote starts in 15 minutes. We'll see how different it is from last year. &#59;)

The WiFi is free, the coffee is hot, and the Open Source Showcase room is filling up and the booths are already active.

We'll try to keep you up to date as it happens. But we're not going to try and live up to Coté's calling us a "blogging machine" last year. &#59;D

Thoughts on Open Source Think Tank 2008

The third annual Open Source Think Tank, hosted by the Olliance Group and DLA Piper, will be February 7 - 9 this year, at the Silverado in Napa Valley, CA. As the title states, this is a think tank, and not a conference. The schedule shows two CIO Panel Discussions, four brainstorming sessions and many hours of networking.

Attendees have been asked to think about topics they would like to see discussed, and here's what Clarise and I have bandied about.

  1. Our contention has always been that the strength of open source lies in flexibility and community. These two qualities make open source more valuable than the closed competition. How can new business models build on these strengths, and move away from licensing, subscriptions, support and other means of pushing "80% of the capabilities for 20% of the cost" as the message?
  2. SaaS is a new way of deploying and delivering software to customers, and open source leads to new ways of development of software and new business models to commercialize this software. Some companies make use of one and make money off the other, in various ways and various combinations. Mashups and SOA and "Platform as a Service" are leading to even greater re-use and innovation. What new synergies and business models can come about from these concepts?
  3. Open source software, and even hardware designs, cover many areas of business and consumer applications: operating systems, web servers, databases, reporting, productivity, web browsers, and on and on. However, there are still holes, actual and perceived. These holes exist in segments, such as Governance, in functions within an application, and in vertical/niche markets. Will these holes be filled by existing open source companies expanding their offerings, or by new companies? How does the consolidation, acquisition by proprietary companies, and leveling-off of VC funding for open source companies affect the expansion of open source to fill these holes?
  4. There is continuing resistance to open source in organizations of all sizes. From some smaller companies, we've seen two types of responses: one being that they don't have the skills to use, let alone leverage, open source software, and the other being that they gain credibility when they tell their customers that they are using Oracle, WebMethods, Weblogic, Sharepoint and other "brand" names, and that they are afraid of losing this credibility if open source is anywhere except deep under the covers. From large companies, the refusal is often more abrupt and more definite. Arguments of flexibility, security and lower cost often fall on deaf ears.
  5. Some in the open source world have blogged about the dearth of executive talent for open source companies. What talent and background make for a good open source corporate leader? At what point in an open source company's life cycle is general business acumen more important than open source commitment?
  6. It can often be difficult to find the open source project needle in the haystack of the web. It can be very frustrating to find a project that looks like a solution, only to discover that after several years, it never left the "proposal" or alpha stage. Using tagging, ranking and other Web2.0 social networking type of tools on Sourceforge.net and other forges, foundries and code repositories might help to promote and strengthen ideas and projects into the mainstream, or at least the "Long Tail".

Here are some other attendees' thoughts from their blog posts:

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

Search

  XML Feeds