IBM i Execs Put Database On The Map
May 23, 2016 Dan Burger
The IBM i roadmap goes nowhere without DB2 for i. It’s the database that drives the bus. With each release of the IBM i operating system and each of the Technology Refreshes that come between the releases, we find database enhancements prominent. And this isn’t going to change. At least for the foreseeable future, it’s a data-centric world.
It’s not fair to say IBM i innovation is nothing but database. There’s a lot going on in other areas, but the success of the system has always been the integrated database and the future success of the system will absolutely depend on database innovation. Big data and analytics are reshaping enterprise computing, but that’s not a universal statement that applies to all organizations. Building better business solutions is still the name of the game and solutions have never been universally agreed upon.
Last week, at the COMMON Annual Meeting and Exposition, four of IBM‘s top executives dedicated to IBM i talked database with IT Jungle. Here’s what Chief Architect Steve Will, DB2 for i Business Architect Scott Forstie, Business Architect for Application Development Tim Rowe, and Product Offerings Manager Alison Butterill had to say on database innovation and preservation.
IT Jungle: How does the DB2 for i database and a data-centric architecture fit into the IBM i roadmap?
Steve Will: The key to recognizing our strategy around data is creating a data-centric architecture that allows customers to define at the database the necessary attributes for handling the data. In the past few releases, as we’ve done XML, RCAC (row and column access control), and temporal support, we are trying to solve business data problems at the database so that people don’t waste time doing it their own special way outside of the database.
That assures that key data, when a customer wants to start taking advantage of analytics, we gotten them used to the idea of keeping it on the IBM i, where it’s doing transactions. Now, as we do things like OLAP support, we are putting the functions where the data is.
We see the data that has always been important to transaction processing being key to a lot of the new workloads we expect will come in the future.
We are trying to get newer technology that deals with data to be centered on not grabbing JSON data from every other place, but to get what is connected to DB2. I suspect there will be more technologies that we want to center on our database.
Tim Rowe: A lot of our data-centric strategy is part of our modernization approach. We are trying to ensure the technologies on our platform are right for not only today, but will be right for the future. It fits together with the other modernization pieces like RPG to the future.
Steve Will: Because this platform is integrated with the database built in, we think about data-related technologies. I don’t know any other platform that would have invested in the kind of services we have and keep rolling out. We are not just a system that needs to provide services, we are a database that needs to provide services.
Scott Forstie: I’m not aware of any other database that is providing the kind of services we have. Our services use SQL and the power of a query engine for a purpose. Are the clients interested? Will they use them? Will they see the value in them? The final chapter hasn’t been written yet, but our feedback has been positive and people see how it can help their businesses.
Tim Rowe: It’s been a requirement for ages that people wanted REST APIs. A REST API provides data. Then you have to write more programming to process and do something with that data. The SQL services harness SQL at the point of the data. It makes them more powerful than if we had just stopped with REST APIs.
Scott Forstie: Emerging technologies go beyond data-centric and going beyond APIs and commands. We have Marc Anderson, our distinguished engineer and chief architect, who spends every day working with the SQL Language Council and the DB2 family, trying to identify things clients don’t know to ask for. An example of this was JSON. Everyone knows what it is, but how do you interrelate it with the database? We are working on it. We delivered a DB2 JSON store and we are continuing to look at this. The SQL language is being enhanced. We should be thinking about it. A more near-term enhancement is generated column support.
We have generated columns that are the engine behind temporal working. We didn’t just stop after adding three time stamps. We gave customers other things to give them row-level auditing. They can do this whether they are in DDS or SQL.
IT Jungle: Database innovation is an ongoing process. What can you comment on that relates to what’s been delivered and what’s to come?
Alison Butterill: Advisory councils are relied on to provide info on what they need to solve business problems. I would say that is the number one driver. Certainly we know of things going on in the industry, but we won’t prioritize those over what our customers and ISVs are asking us for. Temporal support is a good example.
Steve Will: Two of the functions that were enabled with XML–a way to describe and store data–included a way to consume and publish that stuff. JSON is another way to describe and store data. But we need a way to consume and publish it. That’s a natural thing for us to do in the future.
We just had our ISV Advisory Council meeting May 21. Scott Forstie talked about where we might go with JSON. The council members gave some strong opinions about which things were most important. This helps determine what next steps to take and in which order to take them. We have all of our advisory council meetings coming up in a short time, which helps us with the timing of our plans.
IT Jungle: Is it possible for feedback to be turned into action right away?
Steve Will: Depending on how big the request is, we may have it ready for the next TR or we may start working on it right away because it’s important and will take more time than having it ready at the next TR. It may be more important than something easier that could be delivered quicker.
Alison Butterill: We do see things come from the advisory councils that are released at the next TR.
Tim Rowe: Scott and I have both gotten requirements that can be completed in several days, but sometimes it takes three years.
IT Jungle: Some of the requirements come from companies that are early adopters. They ask for it. You give it to them. They use it. But much of this remains unused by the majority of IBM i shops for years. How does that factor into the development process?
Steve Will: The value of some new technologies is not widely known, unless you are a technical person, but services are more easily understood. Once you get the idea of what a service can do, you can adopt a bunch of them. Services allow us to respond very quickly to requests from customers.
When we come out with a major release, there are some key database functions like RCAC and temporal support that are foundational. You might want to upgrade your OS because of a specific function. The open source stuff gets delivered over several OS releases. The message is tied to continuous delivery rather than a specific release.
IT Jungle: How does the database road map correspond with IBM’s overall view of analytics?
Scott Forstie: We’ve been talking with customers not just about IBM’s overall strategy, but about what they (the customers) value, what products they use–whether it’s DB2 Web Query or another analytical product–and how they do operational analytics. Not everybody cares about. Not everybody is doing it. In 7.3 we added a bunch of new OLAP support. Now we need to tell the story, paint some of the analysis picture. For some people it’s easy and they are excited.
Steve Will: IBM’s direction on analytics is a company direction. Then we, within the OS, have to ask how much is applicable to our customers right now and how much is applicable in the future. And then do we need to implement more than what is going to naturally come out of the database roadmap. Like consuming JSON, for instance.
IBM’s direction on analytics is important for us to understand, but we always temper that with what we think our customers need and want and consume.
Certainly there are analytics things coming in the future. We are determining which will be important to our i clients based on what they say. We are not in the business, for instance, of competing with Watson. We are working on analytics on structured business data. That’s where our customers can get value. But we look to the future and say ‘When might we want some of that other stuff.’
Alison Butterill: We have clients today that use the power of Watson. We provide that capability to connect to Watson without having to have a Watson implementation.
There are certainly a lot of products that do analytics in a different way than we do with IBM i.
Steve Will: Different parts of IBM are going to talk about things differently. If you are selling an analytics thing that runs somewhere other than i, you will talk about that.
We see a need from our customer base to do analytics on the same partition they do data processing on. That’s why we are investing in the kind of analytics that the OLAP support does. Because that’s a bridge too far for a lot of our customers. There are real disadvantages sometimes to taking a copy of the data and doing analytics somewhere else. Not only is it operationally harder because of two environments, but you are also dealing with stale data.
Our customers would prefer to have one integrated place to do this. We are not trying to compete with the analytics that run on another platform. We try to make sure that analytics engines that want to get at our data now get it live. The OLAP functions are not about taking the data somewhere else.
I don’t think that is contrary to IBM’s strategy. It is supporting the strategy of doing analytics the way our customers want, which is on the box.
Alison Butterill: A lot of our customers will go to a Linux partition on the same box. Some will go off the box to do analytics. We are trying to provide choices that appeal to the widest variety of clients. We do tie into the products that tie into Linux as part of the choices.
Tim Rowe: We have to provide choice because we have a diverse customer base. They each do things their way.
Steve Will: I think this is a natural evolution of what this platform has been. New technologies come out. Sometimes those technologies are immediately integrated into what we do. And sometimes they grow up on some other platform. When it becomes clear it is something our customer base will want, we find a way to integrate that value.
Analytics started as companies said give me copies of all your data and I will do some analysis on it. But that’s not what analytics is. It’s not about taking data from somewhere, it’s about answering business questions. We think that is a key technology here and in the future. That’s what IBM is saying.
When all you could do was copy data and put it in a repository, that’s what we did. But now we are integrating it because that’s what we think customers will benefit from.
IT Jungle: Many of the database enhancements are coming in pieces with the big picture somewhere in the future. The delivered piece, in and of itself, is not all that useful. How do you view this?
Steve Will: So without totally agreeing with you, let me agree with a basic premise. We built a little bit of analytics before 7.3. But when we decided 7.3 was going to be an OLAP release, then we selected a group of things that our analytics experts said, “You can do something real with that group.” Once you’ve done that, people will still say, “Here’s the next one and the next one.” But we went from having a few to having a large chunk. That large chunk was defined by the major things you need to do in analytics right now.
Scott Forstie: I wouldn’t portray the customers as previously doing no analytical analysis, but the burden was on them to either incorporate packages or write applications that did multi-step processing and copy data in many places because you couldn’t ask the database just a single question. It was set-at-a-time processing. At 7.3, there’s the capability to have single queries that give the final results.
Steve Will: For those getting into analytics we have simplified it and made it faster for them to get their answer. For those not on analytics yet, I think this major release with significant new function will make them say, “Maybe I should be doing that.” And when they start adopting it, they will get the benefit of it being integrated.
Tim Rowe: Our platform has always been a continual evolution. We never deliver the final solution because the industry continues to change. We are always evolving the database, the application development, and the system management. All these tools are a continual evolution to ensure we are future thinking.