Neil McGovern, Senior Director, Product Marketing, SAP Data Warehousing, joins SAPinsider Studio at the 2016 BI-HANA-IoT event to discuss agile analytics and the modern data warehouse. Topics of this discussion include the logical, modern data warehouse, the importance and benefits of avoiding data replication in the data mart, and the role of SAP S/4HANA in the evolution of the traditional data warehouse.
This is an edited version of the transcript:
Ken Murphy, SAPinsider: Hi, this is Ken Murphy with SAPinsider, and I’m at the SAPinsider BI-HANA-IoT event in Las Vegas. This afternoon I’m pleased to be joined by Neil McGovern, who is the Senior Director of Product Marketing, SAP Data Warehousing. Neil is here today to talk to us about agile analytics and data warehousing. Neil, thanks for being with us.
Neil McGovern, SAP: A pleasure.
Ken: I was hoping to start with a definition of agile analytics. What is it?
Neil: Agile analytics is the ability to be more flexible about the way that your users can access data; there’s agility in the sort of slicing and dicing aspect. There’s also agility from a more latency perspective, being more agile and reacting in a more agile fashion to changes in the market. So lower latencies of data, so instead of acting on data that’s a day old or sometimes even a week old, you’re acting on real-time data. Sometimes it’ s mixed up with agile development which is a development process but the two are very distinct.
Ken: Are there other misconceptions about what it is, or is not?
Neil: To that point, we conducted a worldwide Forrester survey on agile analytics and the adoption and so on, and there’s a white paper we have on that that goes into detail on what it is and how it’s being adopted and so on. But we’re seeing more and more organizations adopting it, and one of the findings from this Forrester survey was that faster growing organizations are adopting agile analytics, they’re more real-time, their users are more involved with the analytics process. The users have more ability to access the data and get to the raw data, slice and dice the data and so on.
Ken: So how does agile analytics differ then from operational analytics and why do you see an increased focus in agile analytics for today’s enterprise?
Neil: The first part is, operational vs. agile. Operational analytics is a type of analytics where you’re focused on the operations of your business; so it’s a category of analytics. Agile analytics tends to be more about flexibility, what-if, being able to dive into the data and look around and figure it out. So the two overlap and both are similarly focused on real-time and responding quickly, but operational analytics tends to be about your day-to-day operations whereas agile analytics may answer questions that don’t fall into that day-to-day requirement. There is an overlap.
Ken: We hear a lot today about an increase in self-service reporting. If that’s going to lead to a number of customized data marts, how does a company avoid too much fragmentation?
Neil: Yeah, great question. It ties into a little bit from your last question about adoption. So we’re seeing a lot of organizations adopt agile analytics. The Forrester survey saw that, for companies that are growing at 15% per year or more, over 90% of those have an agile capability. There’s always a price to pay – when we look at the vision for the modern data warehouse, you’ll see where we’re going and when we get to the modern data warehouse we’ll have both agility and we’ll defragment our analytics environment. But in the interim in many circumstances there might be a fragmentation price to pay to get the agility. The reason for that is that – it’s the old analogy that you can’t turn an oil tanker on a dime – and in this case the oil tankers are these big iron enterprise data warehouses. They can’t become agile, but you can bring agility in by bringing in a data warehouse, the price you pay as you said is the fragmentation. To minimize that, what we’ve tried to do is introduce the concept of the logical data warehouse or a data fabric where the data mart itself needn’t replicate the data from other sources. It can reach out and pool in the latest copy of the data to try and minimize that replication and yet another copy of the data being held. So there are techniques to minimize the fragmentation, but we’re seeing that companies that are growing fast and leading their markets are willing to adopt an agile approach and pay the price even if that is with fragmentation.
Ken: And can you tell the audience more about the SAP vision or approach to the modern data warehouse and how it differs from the traditional data warehousing approach, or even BW powered by SAP HANA?
Neil: The concept behind the modern data warehouse is a different paradigm to analytics, to business analytics. The current paradigm basically says that we’ll extract the data, we’ll put it in a separate storage, we’ll build cubes and views, materialized views, and indexes on that data to improve the performance of the pre-determined queries and then the users can run the queries. You end up with – and this is where the lack of agility comes in of course – because the underlying architecture is basically disc-run and it’s very difficult to slice and dice. If you go outside the pre-built cubes, your performance tends to (suffer). The approach for the modern data warehouse is based around the concept that in-memory databases are so fast that they can handle running your applications and running your analytics on the same data. So there’s no data movement, it’s the same dataset. And that means there’s no movement of data, there’s no introducing latency between the data being created in your application and then replicated an hour later or a day later into your data warehouse. So that’s the first part, which is reducing numbers of copies of data, and that’s why I said that the modern data warehouse will address some of the fragmentation, same with agile data marts. The other concepts are, because again with the performance of an in-memory database, you don’t need to pre-build indexes, you don’t need to materialized views, you don’t need to create cubes, aggregates, and so on. You can go straight to the atomic data, crunch that data and get your answer. That is the vision: one copy of the data, working on atomic data, and zero latency because you’re using the same data for both analytics and transactions. And that’s the modern data warehouse.
Ken: For the company that is running SAP BW powered by SAP HANA that wants to get to that modern data warehouse state, what are the integration scenarios? Does that mean migrating to SAP S/4HANA?
Neil: I personally strongly recommend that you move to BW on HANA, it still is older paradigm where you have a separate copy of the data in HANA however the vision is that we integrate those. S/4HANA is not just going to be a game-changer for applications, it’ s going to be a game-changer for data warehousing as well because we’re taking operational reporting that’s very often done in your enterprise data warehouse, moving it into the application. So we’re starting to unload workloads from data warehouse and put it into the application. The goal for BW is that it enables you to further unload a workload from the enterprise data warehouse into BW, but BW will eventually use the same dataset as S/4HANA so S/4HANA will handle the operational reporting as part of its role as an application, but S/4HANA isn’t the data warehouse, BW will be so BW will give you that smart data integration so that you can connect to other data sources and pull fresh data in. It will give you a range of other data warehouse services that S/4HANA wouldn’t be anticipated to provide. And it will also be able to bring in third-party data, handle streaming data, and so on. So there’s a lot that – and I’m talking into the future – but I’d recommend that a customer who is on Suite on HANA and BW on HANA is primed to move to that data singularity, where instead of having two copies of the data they’re moved to one copy of the data and merging the application and the analytics into a single dataset. It’s going to be a very interesting few years.
Ken: Neil, thanks for joining us today.
Neil: Thanks Ken, I appreciate it.
The Forrester Consulting white paper that McGovern references, published in December, 2015, polled 275 global IT and business decision makers responsible for BI to evaluate the current state of BI environments across industries.