Archive for the ‘Conferences’ Category:
Mastering SAP Technology 2008 Conference
I’ve just got home from this years Mastering SAP Technologies conference. Over 50 presentations and 4 workshops in the three formal days, plus a useful SAP Community Day / ‘unconference’ on the Sunday immediately prior to the conference. As usual at these things, the informal networking was as much fun as the rest of the conference (thanks to Microsoft for sponsoring the drinks on Sunday night). One of the highlights was meeting and talking to Thomas Jung and Peter McNulty from SAP Labs in Paolo Alto, USA. There was a lot to learn, and I was impressed by how willing people in general (not just the presenters, SAP staff or vendor staff) were to share their experiences.
Mike Nicholls (talking) and myself (in one of THOSE shirts) facilitated one of the SDN BOF sessions – this one was tentatively titled “Portals implementation technical how to”.
Most of the people who sat in on this did not have an SAP portal at all, and were only considering it because they wanted to implement ESS / MSS.
This lead to some people discussing how it was theoretically possible to run ESS or MSS (or write your own equivalent) without the SAP portal. However, Mike did a good job of explaining exactly how impractical it was.
It was here that I was struck by how underutilised the SAP portal was in most sites (in Australia at least), with little or no use of Collaboration, Knowledge Management etc.
The conference itself was kicked off on Monday morning by a brief keynote from Ishmael Ghalinin (of Intalio of BPMS fame) about living online, a technique called office 2.0. He manages his life through such online tools as Google Docs and Salesforce.com, and in fact, the computer he was presenting with was one that he’d borrowed for the morning. The obvious flaw turned up when he announced that he was leaving immediately to catch a fligh tfrom Brisbane to the US – the direct Qantas flight giving him approximately 13 hours of no connectivity.
Process Integration and Enterprise SOA
Track A (of 5 tracks) was devoted to XI / PI and SOA presentations. One of the more interesting presentations on day 1 came from this track, from Benjamin Salter of Valero. He spoke of taming the Enterprise SOA tangle. The key points, as Mr Salter summarised them were:
Enterprise SOA is a way of thinking -A design paradigm grounded in reusability. Not a “thing”.
It is an evolution. Existing systems don’t change overnight (but we can evolve),
Enterprise SOA requires participation. It is not a spectator sport! It affects all areas of the organization.
Upgrades – Planning, Preparation and Implementation
Track B focused on upgrades and the issues associated with them. A couple of organisations shared their experiences, but the most useful one to me was the “Implementing Solution Manager” presentation from Dion Ellison. Having done installed Solution manager recently myself for a customer, it was interesting to see what differences there were between our philosophies. I had only installed it for some very simple Solution Monitoring. Dion’s customer had also installed Change Management and Test Management.
User Interface – Adobe, NW CE, Portal, etc
Track C focused on the current and new user interfaces.
On day one, Thomas Jung provided a very comprehensive roadmap of SAP’s future UI Strategy. The key takeaway was that the major tool for presenting data to the client is to be WebDynpro. The decision about whether you perform your developments in ABAP or JAVA should be based on what skills you have in house, and of course which particular usage type you are developing for (for example, Portals don’t have an ABAP engine, while PI systems have both).
The other takeaway is that whether content was presented via HTML rendering (i.e. browser of your choice) or the new NetWeaver Business Client would be determined at runtime. The big difference between the two is that a browser will expect HTML and Javascript, as it currently does, with the attendant data transfer size and rendering issues that come with highly structured generated code. On the other hand, the Business Client will accept a smaller file in an XML like format.
On Day two, Mike Nicholls presented on what we could expect from the next major release of the SAP portal. The key takeaways here were that your portal admin and content management skills are still valid, SAP have redesigned the architecture, providing a more robust platform for the portal, and that the end users probably wouldn’t notice the difference. As part of the redesign, the Portal will run in a SAP J2EE 5 supplied JVM, similar to the one provided as part of the NetWeaver 7.1 Composition Environment Java EE Trial Version. We will loose the concept of a J2EE Dispatcher (and therefore the Visual Administrator tool).
On the other hand, We are promised a Newer configuration tool,
more functions in NetWeaver Administrator,
an MMC style administration tools for non-Windows landscapes,
Some Web Dynpro portal administration screens,
newer (and IMHO hopefully more robust) version of NWDS,
and a newer way to deploy PAR files
Infrastructure / BASIS / Security tracks
Track D and E presentations tended to focus on infrastructure, BASIS and Security topics.
One of the more enlightening presentations (for me, anyway) was SAP Workloads in VMwareVi3, presented by Andre Kemp, Sr. Product Marketing Manager of VMWare Asia-Pacific. As part of the presentation, Andre showed a live VMWare ESX server, running on two fairly old physical servers. One of the first things I noticed was how dynamic the changes in resource allocation were; up there with the Workload Management available on larger hardware / operating system combinations (such as AIX 5.x on the larger P5 series machines). Andre showed how, under the same load, manipulation of the resources (CPU and memory allocation) altered the the percentage of utilisation on the SAP VM being altered, and on the SAP workload itself.
In a subsequent demonstration, by Tony Garland of ABB Grain, we saw how you could manipulate the VMs so that ostensibly production hardware could be taken offline with little or no impact, and how the VMWare Disaster Recovery Systems worked, by taking a ‘running dump’ of a live system.
Tim Bohlsen gave a 30 – 40 minute over view of his 2 day workshop. One of the concepts he introduced was Tuning for Response Time versus Tuning for Resource Utilisation. Tuning for Response Time is when you tune your system(s) till you get sub second Dialog response times, and leave it at that. Tuning for Resource Utilisation is when you lighten the load, either by tuning individual components of the response time more and more, or by getting rid of extraneous workloads (i.e running a job once a day instead of once an hour). In effect, as Tim pointed out, removing 50% of the load from a system prior to adding hardware means new hardware is TWICE as valuable. In fact that leads directly to one of the three key points from his presentation –
Even without a crisis, it is always possible to determine the tuning change with the highest potential.
Well tuned system use less hardware, in many cases MUCH LESS !! You can reduce your companies costs through pro-active work.
Squeaky wheels get oiled. BUT…. Shrinking or even removing the wheel can be MUCH more effective.
