Interview with Expressor Software CEO, Bob Potter

We recently spoke with Expressor Software CEO, Bob Potter to find out more about their data integration technology, how it can benefit data migration initiatives and what their plans are for the future.

How long has the company been in existence?

Just over five years now. We founded expressor software in 2003 to address the market need for a higher performing and smarter data integration solution that was both easier to use and more affordable than the legacy offerings.

By 2007, we had completed development of our core parallel data processing engine, so we used the Series A funding we received that year to round out our development team and complete the expressor semantic data integration system.

The product became general release in July 2008 - what has the take-up of expressor been like since that time?

We have been pleased with the take-up we’ve seen so far with customers and partners, despite these challenging times.

Even going back to the beta program before we formally launched in early 2008, feedback has been consistent. Customers have said our channel pricing model is much more economical and flexible than traditional enterprise software licensing schemes, our semantic rationalization approach saves them time and frustration, our role-based development environment makes it much easier to engage business users, and our parallel data processing engine is surprisingly fast and scalable.

Let me give you a real-life example. One of our customers is a major US marketing services firm that is using expressor for an important data migration project, as the company moves off a mainframe platform and onto a cluster of Linux machines.

Of course, we had to meet the performance of their existing vendor – who just happens to be one of the leaders in high-performance DI, by the way. But the two biggest deciding factors were our role-based development environment and pricing model. By making it easier for the data analysts to work with their customers to define business rules and for the technical group to build the actual system, each group could focus on what they do best. The customer described it as each group having their own “sandbox,” where they didn’t have to worry about what any other group was doing.

As for licensing, this customer really appreciated the fact that the expressor pricing model allowed them to sign up for only as many channels as they needed rather than getting locked into an “enterprise” licensing scheme based on theoretical CPU capacity.

Data migration projects are typically time-boxed, i.e. they don't require permanent licensing - can you cater for this in your licensing approach?

Absolutely. To reduce the cost of one-time migration projects, we offer term licenses - for six months, one year or two years. We even publish this on our pricing page on our website..

We believe that companies, particularly in this economic climate, should be able to purchase only the software they need it. That’s why all our pricing is usage-based.

We define usage in terms of channels - which are units of data parallelism. All current and future tools, metadata repositories, and data connectors can be licensed free of charge with the purchase of at least one processor channel.

Customers can run as many applications as they need, as long as no single application uses more than the maximum channels licensed. expressor channels are licensed for a specific server machine, but we allow customers to redeploy channels to another server free of charge by issuing a license key for that machine configuration.

expressor software also offers affordable product maintenance at 20% of the list price for each channel, which includes 24 by 7 premium product support as well as product upgrades.

What features of expressor make the product particularly suited to data migration projects?

I’d have to say it is our semantic rationalization capabilities, which allow companies to rationalize physical metadata constructs to common business definitions and write target-specific data transformation and business rules that are 100% reusable.

Can you explain exactly what you mean by "semantic abstraction model"? How does this benefit a migration/integration initiative?

Sure. We designed the expressor semantic data integration system to solve the most complex, pervasive problems with data integration and data migration - like the “data mapping” nightmare created when building new applications.

Either because of acquisitions, expansion, or even geographical differences, most organizations use different physical metadata to describe the exact same information, such as “customer#” or “customer_no” or even “customer ID," depending on where it’s stored.

So when it comes time to building new applications that tap into these different data sources, they need to somehow reconcile, or “map” all of this metadata to the right business definitions and rules.

This is one of the biggest pain points for traditional data integration solutions, because they require developers to map all of the different metadata directly between sources and targets and build their business/transformation rules directly on top of these physical metadata names. As a result, users have to recreate these mappings and rewrite the corresponding business rules every time they create a new integration or migration job and every time the underlying data sources, metadata or data types change, effectively eliminating any hope of reuse.

expressor uses a highly automated process we call “semantic rationalization” to enable users to correlate any number of different physical metadata descriptions to common business definitions that they can then use in multiple applications.

For example, an expressor user could create a single business definition for “customer_number” that encompasses all of the different physical metadata descriptions I just mentioned.

In addition, users can also define the metatype – or way the data is used in the business, such as text, email address or currency, for each business definition, which means expressor can automatically abstract the datatypes of all physical metadata. This abstraction process isolates the developer from the physical complexities so that they can concentrate on solving the business problem at hand.

Once expressor users have specified their business definitions, the target-specific business rules they create that reference these common definitions are abstracted from the physical metadata - and therefore reusable.

Let’s say your organization has a rule that specifies “if part_number begins with ‘384,’ then ship_type is ‘UPS.’”

In expressor, this rule uses semantic definitions for “part_number” and “ship_type” that can correspond to any number of different physical metadata descriptions, and neither definition needs to have a physical data type or length associated with it.

That way, even if the underlying data source, physical metadata name, or datatype that “part_number” refers to changes, this rule remains unaffected and reusable. Likewise if a new data source becomes available, all previously created rules associated with “part_number” are automatically available for the new source of data.

The final advantage of our approach is that that once customers begin populating their expressor repositories with more and more definitions and rules, the time required for this semantic rationalization process will decrease, since the metadata they need may already exist. We also provide tools to jump-start the process by bulk-loading customer-specific definitions.

As I’m sure you can tell, we are very excited about this and other aspects of our solution – and so are many of the enterprises who have seen and used it. We are also glad that we are the first data integration vendor which is bringing truly new, innovative ideas to the market, several of which we have filed patents for.

You claim that expressor runs in batch or real-time, what are the characteristics of real-time operation - for example does it perform activities such as change data capture or bi-synchronous migration?

That’s a great question.

“Real-time” can mean different things to different people. We like to discuss this in terms of latency requirements. One of our fundamental principles is that we have one engine that will run both scheduled – batch -- and unscheduled , or low-latency, requests.

I assume that scheduled “batch” requests are understood, so let’s briefly discuss low-latency requests.

I use the term low latency because there is a harsh relationship between the latency of an application and the associated cost of the required infrastructure. expressor is focused on delivering solutions for our customers that balance the cost of ownership over time with the business value delivered. “Real-time” is rarely required to satisfy business requirements, so understanding the required latency allows us to solve the business problem while managing the cost and associated risk of the application.

expressor currently supports low latency through the various message queue technologies available, such as Websphere MQ, MSMQ, Tibco, etc. Our architecture supports what we call “service based tasks” – a persistent task that fields requests from various web services sources, like SOAP over HTTP, JMS, etc., performs a task, and optionally returns a result.

These capabilities are intended to support a number of low latency applications, including our cloud computing initiatives and web services functionality. We are currently working with a couple of organizations in delivering low latency solutions – interestingly, cloud computing has been popping up here in the US more and more often. These technologies are powerful but they are at an early lifecycle stage and we understand the rigor that serious organizations require - that is why we are doing the groundwork before any formal announcement.

As for change data capture, we have a number of ways to accomplish this – both within our product as well as by leveraging existing technologies.

Bi-synchronous migration is an intriguing question. Most often this involves database synchronization and although we can play in this space, databases themselves are actually starting to do a respectable job addressing this problem.

You mention that expressor carries out "metadata discovery," data profiling and data correction - can you describe these functions in more detail, is this carried out by a separate product for example?

expressor’s philosophy is that we are providing a complete data integration solution, so there will be no separate products or hidden/add on charges for the expressor system.

Today we provide our customers the ability to integrate many of the popular data quality and data profiling tools into their DI applications. We believe that our vision of semantic rationalization will actually allow us to extend current technologies – not necessarily replacing them, but augmenting their capabilities. We have a number of technologies for which we have patents applied for in this space so I can’t go into details, but I think you can envision the possibilities for profiling and data cleansing to use the knowledge about how data will be used in the business to automate some of these tasks.

Are there any particular markets or solution areas you are currently focusing on that are particularly well-suited to expressor?

Although our solution allows us to address many different DI application areas, we currently target opportunities that take advantage of the remarkable throughput of our parallel data processing engine and our ability to easily handle even the most complex data – including Cobol copybooks and complex XML files.

Our strengths play well in demanding data warehousing and data migration projects. Our smart semantics capabilities appeal to customers embarking on MDM and data governance projects, but we would normally partner with solution companies that have expertise in these applications. In summary, we’re really targeting complex, high-throughput applications at present.

Your marketing materials make reference to the competitive advantage over Ab Initio. This is a bold statement given their achievements and product maturity. Have you had any success in convincing Ab Initio customers to swap platforms?

Yes we have. For example, one of our customers first deployed Ab Initio three years ago. But they never got over the “sticker shock” once it became apparent how expensive their solution was, so they started looking right away for an alternative.

We’re also seeing strong interest from Ab Initio system integration partners, particularly in Europe, who appreciate that expressor software is a much easier company with which to do business.

But the larger point is that it’s not just Ab Initio.

All of the current data integration tools are inefficient and expensive. They’re too complex and labor intensive. They scale with difficulty and are based on obsolete, CPU-based pricing models. By using expressor, companies can cut their total DI costs in half, when compared to current solutions.

In terms of your geographical footprint, which regions do you service now and what's your strategy moving forward?

We have sales and service coverage today in the North America and Europe.

In North America, our direct sales team is augmented by reseller partners, such as Axis Group, as well as systems integration partners like BAAX. In Europe, we currently sell through SI partners such as CME/emergent-I and G7 and will be announcing several other partnerships shortly. We are rapidly adding more partners and see them as a critical part of our go-to-market approach.

What future plans for expressor can you share with our readers?

We have a new version of the product coming out in March of this year that will feature major new capabilities and enhancements in the areas of ease of use, reporting, connectivity -- including some of the low latency capabilities we just discussed -- and even greater throughput.

These enhancements are not only an extension of our original vision, but also the direct result of input from our partners and customers. One of the things our associates have found most refreshing is our intense focus on how we can contribute to our customers’ and partners’ success. I encourage your readers to “watch this space” – we intend to continue to bring an innovative, fresh approach to how data integration is done.

For more information on expressor software:

http://www.expressor-software.com/

http://www.expressor-software.com/resources-overview.htm

http://www.expressor-software.com/products-overview.htm

Recent Interviews:

• Interview with Sami Akbay, GoldenGate
• Interview with Celona CTO Tony Sceales
• Interview with Kognitio CEO Roger Llewellyn
• Technology Interview with Talend COO Fabrice Bonan

Would you like to pose any questions? Please add your comments below.