Here, in the third installment of our series covering SAP TechEd 2016 Las Vegas, Elliott Bajema takes us on a deep dive into SAP BW/4HANA, amongst others things...
Our day jobs never stop and I spend most of the morning on a long client call. Business reality still takes priority even in TechEd and I'm not the only person trying to find a quiet corner for a conference call. In the meantime, fellow Bluefinians Matt Matt, Lloyd Palfrey and Brenton O’Callaghan branch off to take care of their own business – presenting sessions and meeting partners.
I meet up with an SAP contact at lunch and catch up with some of the latest initiatives and programmes of work. I’ve never been keen on the word ‘networking’, but TechEd has honestly proven to be a great place to exchange ideas, meet new people, and plan collaborations. In a rookie mistake, which is probably obvious to veterans of the event, I ran out of business cards on the second day.
Internet of Things
In the afternoon, it’s back to the stand to field questions from SAP customers and partners. As described in my first post of this series, the theme of our display with Lenovo is the Internet of Things (IoT). We are approached by many curious people and it’s worth recounting a couple of the most interesting topics we’ve found ourselves discussing:
- There are plenty of specialised systems from third-party vendors, or even ones developed in-house, which manage sensor data processing. However, if you’re an SAP shop, the exciting part of this concept is the capability to do this on the same platform you’re already using. If you already have HANA for operational reporting, data warehousing, or anything else, you are in possession of a set of tools capable of handling these additional IoT use-cases on a mammoth scale. The resulting landscape simplification this would bring about was an attractive prospect for several people we talked to.
- The use of sensor data in manufacturing lines for predictive maintenance is a big deal. Alongside our ‘oil platform pump monitoring’ scenario, this is the one application of IoT that people really engage with.
- From a technical perspective, people are most interested in the data input layer. We get questions like “How could this possibly support ingestion of data from thousands of sensors, streaming at once? How would the threads handle that incoming data?”. Questions like this are great, and make total sense.
The Smart Data Streaming plugin has been available on HANA since SPS09 and is designed specifically for scenarios like this. Many people weren’t familiar with it.
In addition to looking after our own stand with Lenovo, I spend a lot of time talking to the SAP Data Warehouse team at their kiosk, freshly armed with some curveballs from our recent SAP BW/4HANA Proof of Concept.
Lothar Henkes, Senior Director; Glen Leslie from the Customer Solution Adoption Group; Neil McGovern, Director of Data Warehousing Marketing and the rest of the team are very friendly and accommodating, and bring us up to speed with the latest updates from inside SAP.
Customers are approaching the stand as well while I’m there, and there are some great conversations that ensue. There’s currently plenty of information released in relation to BW/4HANA, including John Appleby's excellent post, the SAP FAQ, and the SAP Technical Overview deck; however, here are some highlights especially for the benefit of those who haven’t been following closely since BW/4HANA was announced back on 9th September 2016.
Front End Tools
BEx is no more. It has ceased to be. Bereft of life, it rests in peace, popped its clogs, shuffled off its mortal coil. It is an ex-tool. We have been referring to this affectionately as ‘BExit’. However, BEx-style Queries are still in use. Now they’re simply called ‘BW Queries’. BEx Query Designer and Web Application Designer (WAD) are no longer relevant, and BW Queries are defined in the BW Modelling Tools in Eclipse. These queries can be run directly in Eclipse if you’re a developer, but there is no Java stack, and they are not intended to be viewed directly in a web browser as many will be used to. Instead, you are supposed to be using specific frontend tools for data consumption. SAP are positioning:
- Analysis for Office for simple analysis;
- Design Studio for Dashboard-style output;
- BusinessObjects Cloud as the cloud-based one stop shop for BI, Planning, Predictive and Sharing.
In addition, generated HANA views on top of BW objects will be accessible with any of the standard SQL tools, no different to native HANA approach to Data Warehousing. Lumira is still useful for data discovery at a business user level, and can be especially versatile if you eschew BW Queries altogether and point it at the underlying aDSO or Composite Provider.
Dynamic Tiering and Data Temperature Control
SAP has invested heavily in this area but for some reason it seems the message isn’t being heard clearly, and there are a lot of people asking about this. It turns out the guidance is straightforward.
- Hot data is kept in the HANA in-memory column store by default;
- Warm data is kept in HANA Extension Nodes (set at the aDSO or partition level);
- Cold data is kept in NLS (both Sybase IQ and Hadoop HDFS are supported as targets).
That’s about it!
Partitioning based on time (latest update time of a record) is performed and controls the routing of data. There is no duplication, and records are simply relocated once the aging criteria are met. The aging parameters are configurable. However, note that right now, this is supported only at partition level; automation on temperature definitions and data placement to Extension Nodes / NLS is only planned for Q2 2017. So if fully-featured Dynamic Tiering is important to you, Q2 is when you’ll get what you need.
It’s also worth taking some time to elaborate on HANA Extension Nodes, as many customers came to the stand with a lack of familiarity here. These are essentially additional standard HANA nodes but with a different sizing formula, access patterns and RAM/CPU ratio. Klaus Nagel from SAP has written some informative posts on them here. From HANA SP12 onwards, and BW 7.5 SP01 onwards, it’s possible to mark certain BW objects as ‘Warm’ when modelling. In this case, all associated database tables are created on HANA Extension Nodes. The advantage is that data storage and cost-effectiveness can be up to quadrupled for an individual node (depending on your deployment and hardware). Partitions are only loaded to memory when necessary. Of course, the tradeoff is that access is slower.
SAP is working on new optimized business content specifically for BW/4HANA and by Q4 this is planned to cover FI, CO, MM, SD and Utilities, with more to come later. Here’s an important thing to consider: if you are planning on going down the ‘conversion’ type migration path, the conversion tool (currently at least) would not, for example, transform an old 0FI dataflow into a new B/4 FI dataflow. So my suggestion is that if you want to make the best use of the simplicity afforded by BW/4HANA, you should embrace the new streamlined business content and take a fresh start, at least with these standard flows.
First of all, seriously consider a fresh, ‘greenfield’ implementation coupled with transportation of converted or compatible objects. If you’re on already using BW on HANA, SAP provide another path: Download the 7.5 add-on ‘BW Edition for HANA’, which includes a tool to perform object conversion (transaction RSB4HTRF), and a program to check for B4H compatibility and perform system conversion into BW/4HANA. From what we have seen, if you have anything more than light customization, you may run into trouble using the conversion tool.
There are promises that the B/4 transformation tool will be improved to convert an entire system and not just a single dataflow at a time. This would include automatic deletion of unsupported objects. This was one of the big pain points we saw when doing this in our PoC and will be a very welcome improvement if it comes to pass. Support for process chain migration should also be added in Q4 this year – more welcome news given that we had to delete all process chains in our PoC.
Interesting information to be sure, but there’s not much time to fully absorb these thoughts as news reaches us that the annual crowd-magnet of DemoJam is about to start! The top six teams from the pre-conference InnoJam event take to the stage to compete in strict 6-minute presentations on their ideas. This year’s focus: Education and Research.
The following plucky teams take to the stage:
- InnoNoobs, presenting its StudyWare app tracking performance and health;
- Techplorer, showing off Gen-E-Us, using sensor data to recommend and plan facility use;
- MisFitbits, also taking the wearable/health approach with Health and Information Tracking System for Students;
- 14 Bis with its Liquid Workforce using Blockchain and AI to approach IP protection;
- SAP Ideas-R-Us is up with Click2Book, combining checkins and seating prediction;
- Keytree, with its Better Education App, BeApp, adding gamification, checkins and a HoloLens.
The audience decide that this year’s winners are team Keytree, and I suspect the fun and flashiness of their Microsoft HoloLens integration is the major deciding factor here as they shoot virtual rockets to Mars (accidentally blowing up Earth as they go).
Heading back from dinner, it’s my brain doing the lion’s share of digesting - sorting through all the new information picked up throughout the day. I make a resolution that tomorrow I’ll dedicate some time to get out and see some more of the expert sessions being held.
Until then, it’s goodnight from TechEd!