What is the relationship between Oracle's SOA tools and products like Hyperion business intelligence BI?
We look at Hyperion and really any BI system as something you want to use to bring intelligence into your business processes that you're using to orchestrate the flow throughout the SOA suite. You also want to see what's happening in your processes and in your BI so you can see where bottlenecks are, so you can optimize your processes. We do a pretty good job of being able to incorporate data from almost any data port into our SOA foundation, so you can use that information as you are reaching decision points in the process. Every process branches somewhere, so the question is how do I know whether to go right or left. Often times, that's a business decision based on information stored in your BI repository.
The other piece of that which we often see is our business activity monitoring piece, which is using a real-time management framework to put a dashboard on top of your business processes. Often times, you want to compare that real-time information with the stored information. So you might say: "It looks like orders are down this morning. Let's look at the same period last month and find out if we were down at that time. Maybe it was just one of those seasonal shifts." You avoid raising red flags unnecessarily by comparing what's happening in real-time with what's stored in BI. How do you get that real-time information?
We tend to pull that information in from the data warehouse in the same in-memory analysis where we're doing all of our real-time aggregation. The challenging part of this is the real-time aggregation of all the process information, event information and transactions coming in all at once. You need to have a special tuned and multi-faceted system that is able to keep up with that processing level. That tends to be in-memory based. Once you have that piece, it's pretty easy to reach out to the non-real-time systems that have historical information. You need to be able to execute a query and have a results set, keep it cached for however long it makes sense and compare it to the real-time information. What is required from business activity monitoring to be able to show executives the value of SOA?
There are two things that are important. First you need to have a BAM front end that is designed for business users. This is different than a systems administration console where IT people are used to looking at dashboards for more technical information. Our view is really intended for business users so the front end for our BAM dashboard reflects that. It's highly visual, highly intuitive. Business users can actually assemble their own reports like you would build a PowerPoint. It's designed so it's as easy to build a real-time dashboard as it is to build a chart in Excel. That is something that makes it accessible to business users.
The second thing is you need to be able to incorporate data from almost any source. We know that business users have pockets of information stored in a variety of systems. If your BAM system is tied only to your processing platform, if it can only accept data from where you're doing your own processing, it's not going to be effective enough. You need to compare data from a variety of systems. So your BAM dashboard needs to have data from the business processes you're executing as well as any other system that's storing information.
Where is Oracle with data services? Do you have the abstraction and integration tools you need right now?
This is exactly where our application development framework (ADF) layer fits in. One of the ideas behind ADF is that any of the data sources can be viewed as a service by the framework. So the developer can bring these data sources together without any real knowledge of how they work. We provide an abstraction layer that will manage the transactional semantics through a data control layer. So the assembly can be handled by the developer in a very simplified way. It hides all the complexity. Do you have anything new coming from your data services roadmap?
I think we have a pretty complete offering for data as a service. We have enterprise service bus technology. If you have non-service-enabled database applications we can put it into a service format and service interface. We also have a data integrator, which we've found useful when you are dealing with high volumes of data. Sometimes a Web services infrastructure just isn't tuned for that much volume so the integrator hooks up heterogeneous data systems and composes it as a service, and doing it in 50 megabyte, 100 megabyte data transformation and loading from one data source to another.