Brox IT-Solutions GmbH

An Interview with Hans-Christian Brockmann

I met brox's engaging founder in a Starbuck's in Louisville. Hans-Christian Brockmann has embraced open source software with verve. The ideas tumbled out of him as I probed about open source search and frameworks like Eclipse. Open source search systems have found purchase in a number of high profile applications. For instance, Yahoo and IBM use Lucene for the companies' free desktop search application. Mr. Brockmann wants to push even further into the enterprise. The full text of the interview appears below.

What is the background of your company?

brox, based in Germany, has been around for 10 years, founded in Nov. 1998. It is notable that we are publishing SMILA at the Eclipse Foundation exactly 10 years after we started working toward our vision of creating a standardized infrastructure for information logistics.

We started brox with the vision that future applications would seamlessly combine structured and unstructured information. We saw that enterprises were gearing up efforts to set up data warehouses in order to get a grip on available structured information. But what value is there in structured information if you are not able to provide a context in which to evaluate this information? Between 70 and 80 percent of enterprise data is actually in what we call an “unstructured” form. This unstructured information contains explanations how processes have to be executed, why decisions were taken and how parameters within the process must be interpreted. So truly this unstructured information holds a lot of value.

We have spent the past 10 years designing an appropriate infrastructure for handling, interpreting and presenting personalized information from vast amounts of data within enterprises. Our key customers are companies like Volkswagen, Audi, Bosch, Credit-Suisse etc.

I recall hearing you say that there were 180,000 Lucene search projects underway. What's the significance of this interest in open source search?

Having spent 10 years catering to enterprise customers, we are very much aware of the strategic and political aspects of deploying IT-infrastructures to large organizations. Our project SMILA (SeMantic Information Logistics Architecture) surely can be considered an infrastructure project which embraces open source search within it. Open source solutions are being absorbed in the enterprise, as evidenced by Lucene interest.

The statistical significance of the 180,000 Lucene projects is it underscores the sheer amount of “do it yourself” projects out there, actually massively more than there are professional commercial search implementations.

We want to provide these “do it yourself” search projects with the tools they all are missing..

In our minds the problem with standardization of IT is that business users do not have any incentive in supporting such an initiative. They are incentivized based on their business units’ financial results. So, they want to use the IT that best fits their business needs. The consequence seen is that they will actually oppose standardization of IT services because they perceive it might be detrimental to their success.

At the same time they do not care how the “plumbing” is being done.

SMILA provides CIOs with the perfect answer:

SMILA allows them to create economies of scale on the part of the semantic infrastructure, while at the same time being fully open to use highly specialized business applications and even technology plug-ins to meet user requirements.

Implementing a semantic infrastructure is a long term investment. Our customers frequently think in time ranges of more than 25 years. An open source project will constantly adjust itself to the needs of the market. A closed source product will always be dependent on the innovative powers of a single vendor. What are the chances that a single vendor’s product will be the “be-all, end-all” solution in 25 years?

We believe that open source is not a threat to the CIO but really the one and only common denominator that will provide him with a truly feasibly longterm option of “putting in the plumbing” for a semantic future within his enterprise.

Other open source search projects demonstrate the user’s willingness to try “free-stuff”. It demonstrates that there is a huge grey market within enterprises with people setting up demos, pilots and live portals simply using free and even unsupported software. CIOs cannot stop the increasing use of open source because existing governance processes use budget restrictions to limit and control the kinds of IT implemented. Simply put, open source is the cheaper option. This is only going to grow with the economic setbacks companies now face globally.

SMILA acknowledges the fact that business users will always want quick fixes. It also acknowledges that CIOs typically have their preferred search vendor. Why not ask those enamored with Lucene to also use SMILA, so that they can re-use existing interfaces from other projects within their organization?

We want to provide them with a strategic perspective.

Using SMILA as their infrastructure these projects will be able to create a faster ROI immediately and will significantly reduce maintenance cost in the medium and long run. It’s time for open source to evolve beyond just search. Infrastructure projects will become hubs and spokes within the SMILA based semantic information logistics infrastructure , networking all relevant data sources.

What about Nutch, Solr, and Flax? Will these see similar interest?

Each of these projects address a certain aspect of what we have in mind for SMILA. However, SMILA is not intended to create search algorithms, linguistic technologies like classification algorithms and entity recognition etc.

We highly respect the work and commitment that has gone into these three projects. Our focus with SMILA and its commercial distribution, eccenca, is to provide CIOs with an enterprise class open source alternative that is in all respects (functions, features, service, support, indemnification) comparable with commercial products. SMILA provides the “feel good” integrated, all inclusive packages for everything one could ever need when working with unstructured information. At the same time the OSGI based distribution building process of Eclipse allows us to ship SMILA with a footprint small enough to run either on a handheld device, or within a cluster, or even grid scenario.

We are not competing with other open source projects, we are trying to build the functionality that is missing to make them a true alternative to existing and future commercial products.

We believe that Nutch and Solr are great tools that create a lot of excitement with people who already have a good bit of experience with enterprise search. In contrast, the majority of those implementing Lucene are not experts, but more often than not, search beginners.

SMILA and particularly eccenca will cater to this audience by reducing the barriers. SMILA is aiming to provide a smooth way into enterprise search for first timers. SMILA will be available for download, including Lucene. It will provide instant gratification to the user because it will walk the user through a number of easy steps to set up their own first index.

We are not competing with Lucene, we are just trying to prevent the kinds of maverick implementations we find with so many of our customers. By making it easy to implement connections and setting up indixes in a standardized way, SMILA will help these first-time users .

Why is open source gaining traction?

We believe that open source is just one more way in which software can be built. Other viable ways include outsourcing, off-shoring, and 4GL. In the end, the customer really does not care how the software has been built, as long as it offers excellent value for a reasonable price.

As a consequence we do not believe that open source is the correct solution for every problem. At the same time it is the ideal solution for some problems.

Putting in the plumbing for the next generation of semantic applications is something every organization will have to do in order to remain competitive. In the course of this they will more or less all stumble across the same issues. To name a few: Security, scalability, connectivity, longevity and of course, maintenance and support for such a large infrastructure.

We suggest it is best to share the cost of implementing and maintaining the infrastructure – which is not part of the strategic competencies of any company – on a shared basis. If you look at the top 1000 companies globally and ask them to deliver a list of application software they are using,the lists will be almost identical. The breadth and depth of use of certain products may be different, but basically the tools are similar.

Open source and particularly the Eclipse Foundation model of open source has proven to be a great way of sharing infrastructure costs within consortia. Look at the cost of implementing and maintaining an IDE. More than 100 companies share the cost of developing the Eclipse IDE and thus have created much traction, innovative power and market reac., Today approximately 70% of all Java developers use Eclipse--actually 90 percent of enterprise level developers.

Do we expect to see that all search algorithms, plug-ins, connectors that are available for SMILA will be open source? No, not at all. The Eclipse IDE model has demonstrated that a combination of providing a shared platform including architecture standards free of charge, in combination with a sound plug-in architecture will work best--if it can be combined with a commercial software offering. We already are seeing that commercial search engine and plug-in providers are providing us with resources .

Where does your firm fit in the open source initiative?

At this time we are investing heavily into the creation of the SMILA platform and are currently staffing approximately 70% of the 20 full time developers on the project.

We believe that open source software can only have a place in the enterprise if there is indemnification, 24/7 support, service lever agreements, training and consulting services.

brox will provide a commercial distribution of SMILA under the name eccenca. For eccenca, we will offer all the above mentioned services including migration from SMILA to eccenca, certification and warranties to the customer. In addition we will attempt to play an active roll in creating and sustaining a lively eco-system. We already have many solution partners and plug-ins. One new plug-in may create a dozen new business opportunities for our solution partners. The same is true for the eccenca enterprise customers. New plug-ins or eccenca-based solutions may create new business opportunities for them as well. creating.

What products and services does your company offer?

We are offering an enterprise class commercial distribution of SMILA under the name eccenca.

This offering will come with all the bells and whistles known from similar business models executed by companies like Red Hat.

Beyond the distribution and the included indemnification, support etc. we will offer training, consulting, certification and a knowledge-sharing community.

How can a company using open source protect its intellectual property?

The Eclipse Public License (EPL) is the basis for open source projects within the Eclipse Foundation. More than 200 commercial products of, for instance, IBM are based on core components developed and published under the EPL.

The brox core competency is within our commercial offering around the open source initiative. Our core competency is in enabling the ecosystem to maximize the value from the open architecture standard created within the SMILA project.

Those companies using SMILA based on the EPL will not be required to disclose that SMILA is included within their product. They also have no limitations as to the license they want to apply to their product or solution that integrates SMILA. Thus the fact that SMILA was created in an open source setting has NO effect on the IP of our partners, beyond the contributions they have willingly made. To our knowledge it will also have no effect on software usage and the user – consider that 70% of all java code is developed based on the Eclipse IDE published (like SMILA) under the EPL.

What is your view of IBM's success with UIMA?

We have made various attempts to contact IBM – who are strong contributors and founders of the Eclipse Foundation – about cooperation with the SMILA project. To this moment we have received the feedback that the commoditization of semantic infrastructures to the degree of SMILA does not fit IBM’s strategy of selling proprietary products like Websphere, DB2 etc., a position that I can fully understand.

At the same time I cannot comment on the success of UIMA, but I believe that SMILA is to some degree the logical consequence of predecessor projects like Nutch, Solr and UIMA. We see a strong need in the market for one stop shopping. Customers want to have an end-to-end provider for the entire value chain.

The SMILA project is the first and so far only vendor independent, open and fully integrated end to end solutions for enterprise information logistics.

What will you do to achieve a stronger uptake of your connector idea?

Key to our connector strategy are two factors:

• The value of the SMILA/eccenca eco-system
• The simplicity in which connectors can be implemented

We believe that open source is a great way of creating and sharing connectors. One may build a connector for one’s own purposes and publish it at Eclipse. A few months later one may find that another user has taken the connector and expanded or updated some of its features. That way the cost of creating connectors will in total remain the same as with all other projects – but the cost of maintaining the connectors will be shared within the eco-system.

Brox is considering the option of including some of these connectors within the eccenca distribution and thus will most likely offer commercial level support also for connectors.

In practice this may mean, that two or three or more companies who share the need for a connector can contact brox or any other service provider and ask them to build and commercially support this particular connector. The SMILA documentation and roadmap are public, so there are no risks, hidden agendas or surprises.

Where does the Eclipse Foundation come into the enterprise search and information access effort?

The Eclipse Foundation is currently in a significant transition phase. It is very well known for its IDE initiative and had great success in this field.

The Eclipse Foundation has identified multiple vertical and horizontal markets that could benefit from a similar approach. Projects like SOPERA (an open source SOA infrastructure initiative) and SMILA are still focusing on IT infrastructure. Other projects we may see in the future will actually address business scenarios as for instance the automotive electronics tools chain, where manufacturers and suppliers collaborate to establish standards in order to reduce cost. Another industry that is getting traction is the mobile devices industry. Motorola and Nokia for instance have announced a consortium to share development cost and to standardize parts of their infrastructure using the Eclipse Equinox runtime platform.

What can be done to prevent IBM, Google, Microsoft, Oracle, or some other company from making a proprietary solution based in part on open source?

The Eclipse model is designed in a way that aims to support such initiatives. We would welcome these and any other player’s initiatives to support our cause by re-using parts or all of our work.

You see, open source is not about creating value by excluding people. It is about including them, increasing the size of the pie and thus creating value for everyone.

The uptake of SMILA by all parts of the industry is in the best interest of the entire industry. Because SMILA is free of charge anyway, nobody is going to make significant amount of money be pushing a proprietary standard any longer. The value to the industry is to provide significantly better solutions to the customer at a significantly lower cost, by sharing the cost of the infrastructure.

The customer in the long run will not pay anyone for the infrastructure, but he will pay for innovative solutions that have an effect on its bottom line. So we welcome everyone who shares this notion, stops wasting his money on building proprietary infrastructures and starts putting his R&D dollars towards truly innovative customer values.

As you look forward, how will open source companies like Red Hat and Novell evolve?

I have already mentioned the “one stop shop” need we identified in the market. I am convinced that these players will evolve to becoming a “one stop shop” with a complete stack of pretty much everything a CIO might need, in terms of infrastructure and very common enterprise software.

In terms of search and information access, what are the big trends you see emerging in the next nine to 12 months?

Of course I am convinced that SMILA will become something like an insider’s “Innovation of the Year” toolset in 2009. The response we are getting from the product vendors in the market is that due to the economic situation more and more of them are questioning their budgets for proprietary infrastructure developments.

Successful search vendors will continue to “vertica-lize” their applications to better suit specific industries or business scenarios. In addition customers are requesting a smooth integration of semantic information and business intelligence. We are seeing some vendors who are making huge strides in that direction. We anticipate that SMILA will play an integral part in delivering higher value solutions that combine business process, BI, and search information in a new breed of user dashboard.

Arnold IT Comment

A number of companies have built a business around open source. brox is a young company with an energetic and passionate leader. With the economy weakening in North America and the European Union, brox has an opportunity to move open source search and content processing into another dimension. We're rooting for him and his company.

Stephen E. Arnold, December 5, 2008