SoH2oP - and Intel is the loser!

Earlier this year, there was much rejoicing among our worldwide CCC leaders and grassroots contributors. The IBM Power Group saw the light at the end of the tunnel.

It was obvious that Hana on Power (HoP) would eventually become a reality. There have been too many rumors over the past three years to deny the fact that intensive work is being done on a solution.

I myself regularly inquired in Walldorf - unfortunately I have less good connections to Potsdam to the HPI (editor's note: Hasso Plattner Institute at the University of Potsdam).

And in Walldorf it was very difficult to get answers. There were two factors of uncertainty: first, to what extent is the Hana code actually optimized for and dependent on Intel Xeon, and second, after the disaster with IBM DB2 Blu: is there again a reasonable basis for talks between IBM and SAP regarding Hana?

(Editor's note: Blu is the in-memory computing extension for DB2 that is given away for free to SAP/IBM existing customers, which in some cases may be anti-Hana).

Officially, there is HoP for BW. Theoretically and experimentally possible is "Suite on Hana on Power".

One key to the "Hana Enablement" is Linux. Of course, a Linux version for all Intel processors was available from the beginning, but also for many other platforms, so that parallel development of the Hana database would have been possible.

The close partnership between Intel and SAP, as well as Intel's generous support for HPI, was a conceptual requirement.

In addition, there was the secrecy on the part of SAP and Professor Hasso Plattner. Initially, partners such as Intel and Suse Linux were only told about a follow-up project for the BWA (editor's note: BWA is the Business Warehouse Accelerator, a hardware appliance for accelerating SAP BW).

It was only much later that the developers involved learned that this was to become an independent database and subsequently a platform for the next generation of ERP software.

Hasso Plattner and his colleagues at HPI were committed to Intel right from the start. Which also made development much easier and Hana fast.

But this Intel Xeon centering later made adaptation for a virtual environment under VMware more difficult. Demanded by many customers, SAP struggled for a long time to provide a stable Hana version for virtual servers.

The code was revised for the first time at that time. Obviously the adaptation for IBM Power processors was less complex, because I heard nothing about spectacular failures and anomalies.

With the first version of Hana on VMware, we had the phenomenon in our lab that sometimes the virtualization was faster, but also not quite stable and computationally accurate.

Of course, this summer I also received questions regarding a Hana version on Microsoft Windows. Technically, this would also be possible, but it would not make sense from a market policy perspective:

Microsoft cannot have any interest in building another competitor to SQL Server, and SAP also does not have much ambition to promote the Windows Server platform - the relationship with Suse Linux is quite excellent and is even to be expanded.

With OpenStack and other open source projects, they also want to push VMware - similar to Oracle - out of the SAP community. A very ambitious project from Walldorf! But there is currently an OpenStack euphoria.

So I expect 2016 to be a veritable battle for the most efficient data center virtualization (not just servers, but also storage and networking) as well as the cheapest hybrid cloud, because with the release of AWS, Azure and very soon an adequate Google offering, it will be a price war in the cloud.

SAP will intervene massively in the vendor competition with its own ideas and get Suse Linux and the entire open source community on its side - OpenStack is just the beginning!

SAP wants to solve the issue of indirect use less elegantly. At the moment, it looks like numerous individual solutions that are shaped entirely according to personal preferences and strategic thoughts.

The chaos is perfect, because how can a Hybris e-commerce solution be sold when millions of online users are suspected of indirect use.

Ultimately, half the IT world would have to transfer indirect-use licenses to Walldorf because Ariba, SuccessFactors, Hybris, Fieldglass and Concur are now based on Walldorf software - a disastrous affair for users, but also for SAP.