By Susan McNeice, Senior Research Analyst

About the company:  Boston-based BEZ Systems develops advanced analytics software with predictive modeling technology to solve important database and application performance issues, frequently while they are still small and before they can negatively impact end-users.  The results provide IT decision-makers with the time and information to make informed business decisions deploy resources better and ensure that they meet their service-level commitments to their business partners.

Frost & Sullivan (FS): BEZ delivers what it calls “a business lens on IT.”  What does that mean and why is it important?

Dan Haley (DH):  The reason we came up with that it as sort of a moniker is really the market driver behind IT becoming more business-like, more business focused.  When you look at the area that we focus on in terms of performance, specifically predictive performance, there’s a lot of data there that’s organized in different ways, but not a lot of data to help the IT folks better serve the lines of business.  Our job is to take the existing performance data they have and align it in ways that make sense for people to understand service levels to the lines of business. 

We put our lens on performance data so it’s not just abstract response levels by server or other things that don’t make sense to a business person.  It describes performance in terms of response time, throughput and utilization of resources by lines of business so you can more explicitly serve those lines of business. This gives IT executives the ability to sit with a business partner and develop realistic Service Level Agreements (SLAs).  If they don’t have this, it becomes impossible to put systems in to see if IT is meeting its objectives.

FS: What are the key events in BEZ’s history?

DH: The company was founded in 1983 by Boris Zibitsker.  He brought with him some key technology around predictive algorithms.  He created a successful consulting firm, and like many consulting firms, he eventually decided to create a product around that success.

In 1996, he started developing products for the Teradata market.

He raised venture money around 2000-2001 and he further developed a product called BEZ Plus.

At the end of 2003, we decided on a different approach to the product.  Originally it required a highly trained and somewhat sophisticated user, sort of a “prediction guru” and you really had to understand in detail how you wanted to drive it and what you wanted it to do. 

But that isn’t the way the world works today.  Products today need to provide information at a glance, push information to the user and be as automated as possible. So we ripped the algorithms out of the old product, rebuilt it on a Java-based platform, and designed it in such a way that requires very little interaction between system and user unless there is a condition that requires intervention. Even that is automated so that the system pushes out a message to you.  It is also “wizard” based throughout.

FS: So now that the product has been rebuilt, how does it add value to the CIO’s life?

DH:  For BEZ Prophet, you set up your system and input your performance goals such as the acceptable response times, throughput and utilization for an application server and database.  The system monitors actual performance against those goals and when it sees a condition that triggers a threshold to be exceeded, it alerts you. Once you get into the system, you can quickly drill in and find the potential source of the problem and the system actually offers advice on how to fix the problem.

Then when you add our predictive capability, we take it up another level.  We have a “prediction engine” in our product that allows us to forecast what a system will do in terms of response time, throughput and utilization. We  can not only see if we’re meeting service levels, but we can predict if the system is going to continue to meet the terms of the agreement, and we can take remedial action before it becomes a problem.

As an example, when the system forecasts a breach in a required service level, it will alert you. Let’s say it will tell you that in three months the response time in the order processor line of business will likely drift above the guaranteed two-second response time.  Then it also tells you the likely places where the problem exists.  This is important because performance problems aren’t always directly with your line of business.  It can actually be another line of business that is growing more quickly than anticipated and so it is consuming more system resources than it is supposed to.  There are a number of reasons why this happens, but we quickly allow you to figure that out.

These capabilities are inherent in both BEZ Prophet for databases as well as our newest product, BEZ Vision, for applications. Initially it supports Oracle products and then we’ll release it for other products such as App Server and Teradata.

FS:  BEZ develops products to assess and predict performance in commercially available database products. Why haven’t the database companies already developed a product like yours?

DH: It’s interesting; every time I do a startup I ask myself that question.  To do this job without the software, you need a really unique skill set.  You have to be more than just a brilliant mathematician. There are lots of PhDs out there who understand queuing theory, which we use. They also have to understand databases, application servers and optimization theory.  Not very many people can do that.  We’ve got a research department of six people who do only this kind of thing. In a company of our size, you wouldn’t expect to see a research department, but we do, because the nature of the product requires it.

FS:  What differentiates BEZ from other industry participants?

DH:  Our core competency is in our predictive capability; it’s a unique skill.  Other products have modeling software along the lines of what I would call classic capacity planning, which are really occasional-use products.  We’re not looking at database servers and application servers as “black boxes.”  We get inside and look at how the code is organized, see the workloads and do much more detail in terms of prediction.

We turn the performance paradigm upside down; we look at it in terms of business goals and not just system capacity planning.  Sure, we do the classic metrics of server utilization, and the like, and if we see a threshold that’s going to be exceeded, we’ll look at in detail and see if capacity is a problem.  In that case we do have a capacity planning function, and it can even be used on an ad hoc basis to do “what ifs” on capacity.

It’s the SLAs to the business that drive our system.  Here’s an example.  If a customer has an order entry system, and it’s critical to the business, they might say they need a two-second response time.  Then we go out and collect data and monitor system performance.  It can tell the IT people that at the rate of business growth and system usage they’re going to exceed the two-second SLA in six months.  Then to keep that two-second response, the system tells them what they have to do in terms of re-organizing the system and what they’ll have to buy and when.

It gives them a complete document that summarizes the assumptions, describes performance, provides information on the usage of the system, and recommendations on what to do.  In this case it may be a $500K expenditure, but the IT manager can offer alternatives, so if the business manager is willing to change to a 2 ½ second response time, they won’t have to buy new equipment.  What this means is they’re having a business discussion around meeting service goals as opposed to technology.

FS: Your customers come from information-intensive businesses such as retailing and financial services. Are there other vertical markets you would rank as ideal candidates for your products? 

DH:  Wherever IT - or “BT” for “business technology” as some are calling it - is a critical part of the business.  This includes banking services, stock transactions or selling services and products.  We’ve been getting a lot of interest from the healthcare market and we do have some telecom accounts, again where information is critical to delivery of services and products.  The hosted software market is also an ideal candidate for us because they’re all about service levels.

FS:  Many of your customers are North American companies.  In what other markets is BEZ found, and which markets are next?

DH:  We have distributors in the UK and Brazil, but we wanted to build up our North American base before we headed off in too many other directions.  We’re also talking to people in Japan, Germany, France and Italy - obviously the bigger markets.

FS:  What strategies does BEZ use to strengthen its position in key markets?

DH:  We work with distributors, but only those who understand performance in IT systems. We are talking to larger systems vendors, too.  We also work with the vendors whose products we work with such as the application server vendors and database vendors. 

FS:  How do you position your product to your prospects, and how is this strategy different from others in your space?

DH:  This can be an ROI-based sale, either on revenue gained or cost reduced. We find people think the product is intrinsically valuable or it isn’t.  We do everything by trial; the first thirty are free and they can see the benefits for themselves. 

Fifty percent of the accounts “get it” right away and they realize this will ultimately help them in terms of their revenue goals.  Now some want to talk about it in more detailed terms, and we have ROI models they can use, but it all has to do with the impact of system processing.  For example, what does it cost you in terms of meeting your revenue goals if your web server is down for an hour and you generally process $1 million dollars per day through that channel?

There’s also a cost-based view, that is, what’s the impact of being able to defer capital purchases because you better understand your needs.  One customer actually used our product in modeling a server consolidation project, and justified the purchase of our software with the savings from the consolidation.

With the advent of products that do data center management and automatic provisioning of resources, a lot of people also come to us and say they want us to identify the performance impact of adding more hardware and identify potential performance problems before they add the hardware, so we find ourselves “front ending” these products, like BladeLogic, quite a bit.

FS:  Today’s IT environment is a far cry from the old EDP or MIS departments of the past.  CIOs are expected to be business and technology savvy. Which technology and business developments work most in BEZ’s favor? 

DH:  One of the biggest things IT has to do is manage change.  Their environments are incredibly sophisticated and complicated, yet they have service objectives to meet and many things are changing, all at the same time:  business rules, business goals, technology.  The CIOs are in the middle trying to make senses of all of this, wondering where to place their bets. 

There are a lot of products that help with that, but we fit right in the vortex of that change.  We can help them with all the “what ifs” in advance, if they’re looking at big performance-affecting decisions, say, moving from BEA application servers to IBM WebSphere, or from Oracle to Teradata, for example. We can quantify the impact of their decisions.

At a purely technical level, Services Oriented Architecture (SOA) and virtualization are also big challenges and opportunities. We’re well suited to the way IT builds software today with the emphasis on SOA. Our whole architecture is built with SOA in mind, so we’re built to fit into that perfectly.

We’ve also been working on virtualized environments.  We handle a certain amount of that today, and in the next two years we’ll be doing a lot more in that space.  Virtualization provides a clear financial return; the math is really compelling if you can do it correctly, but we’re still at the beginning of that.  It’s not really in the database server market, and it’s just getting into the application server market, and the issues around performance are going to be critical, so we think that’s a great market driver for us as time goes on.

FS:  Are you ever concerned that IBM, Oracle and the others – most of whom have expertise in predictive analytics - might develop their own version of your products?

DH:  No, we’re not.  Yes, these companies have predictive analytics capabilities and products, but they are along the lines of marketing and customer analytics.  Predicting performance is different from predicting customer buying habits; you need a completely different set of skills to produce forecasts for systems and databases, so no, we don’t think that’s a particular threat.

FS:  What’s next for BEZ? 

DH:  Tactically speaking, we expect to work with new database vendors, but our technical vision is to cover the entire application “stack.”  Today we cover databases, and as I mentioned before, we are releasing our new product, BEZ Vision, for applications.  Then we intend to move up the stack into the middleware tier, predicting performance for products such as [IBM WebSphere] MQ and Tibco, and also the virtualization space. 

You’ll see some business development announcements in the first quarter of next year with respect to new markets.

FS:  What role should we expect BEZ to play in shaping the future of IT?

DH:  The most important issue is managing change.  CIOs are sitting between the lines of business and all this technology that tends to roll over every eighteen months to two years. It used to be you put a mainframe in place and it stayed there.  That’s no longer the case, and I think the CIO’s ability to really understand the business needs of the company and have people who know enough about where all the technology is going to make the right bets so these two areas don’t get out of synch is the place where we’d like to be – this is our focal point.