Some of Nis Sorensen's IT Projects
Here you find link to my CV on Linkedin as well as description of several project done by me directly or by team members under my management.


Curriculum Vitae

You find my resumé here.


Articles, 2009
In the area of Data Centre powers saving, I have written a few articles / opinions.

Published an article in Danish weekly for engineers "Ingeniøren" describing how to use gravity to enhance cooling of racks in cold/hot aisle setup, (in Danish): Tyngdekraften kan spare penge.

The following article in Danish, gives advice in how to reduce power consumption in Data Centres. Particular advice is given in how to make cooling more effective, Nedsæt driftsomkostninger af dine servere (pdf, 100KB).

AES, American energy supplier, 2006
SAP IS-U, Oracle RAC on Linux

Situation:   The customer wanted to move their handling of meter r eadings from a legacy system to SAP IS-U. Their strategy was to move to RedHat Linux 64-bit. Customer wanted a Proof-of-Concept showing how many industry standard servers they needed for handling 1.000.000 meter readings and invoices in three hours and how well that could be scaled to triple the readings in three hours.

Solution:   Piloted Oracle Cluster File System v2 and Oracle RAC in 64-bit version for RedHat. Two 4-socket HP ProLiant servers as Oracle RAC database servers were installed along with the needed amount of application servers for handling the requested amount of meter readings. When IS-U meter reading, invoice calculation and invoice generation were tuned, the time were measured to below the three hours requirement. The amount of database and application servers was increased to be able to handle three times the amount of meter readings in three hours. The scaling showed an expected 80% scalability.

A system handling around 300.000 invoices daily has been implemented at AES Eletropaulo. This system is achieving excellence, in its innovative SAP landscape, multiple application and data base servers using Oracle RAC and Linux, ensuring high availability environment, data integrity and reliability for the business areas that now run ERP, Billing and Commercial business processes in a fully integrated SAP environment.

Pirelli, Italy, 2006
Pirelli implement SAP with Oracle RAC and HP Serviceguide for adaptive computing

Situation:   Pirelli want to have a more flexible SAP system using cluster technology for dynamically indrease or descrease the compute power on both their Oracle database system as well as on their SAP application servers. Additionally they want to have the Enqueue Server protected by using SAP's Replicated Enqueue functionality.

Solution:   Pirelli, a longtime HP cutomer enhanced their HP-UX system with a clustered system for SAP using Oracle Oracle Real Application Cluster protected by HP Serviceguard Extension for SAP. The installation was the first pilot system on HP-UX with SAP on Oracle RAC managed by Serviceguard. Serviceguard Extension for SAP was developed in team managed by me.

Result:   Pirelli got a very flexible adaptive system, where compute power can be increased or decreased on demand using HP Serviceguard Extension for SAP.

HP success story: Pirelli combines HP, Oracle and SAP technologies to protect vital business continuity (pdf, 300KB)

Read the Oracle Success Story: Pirelli Places its SAP Applications on a New Foundation with Oracle Real Application Clusters (RAC) (pdf, 115KB)


HP Competent Cluster Service, 2005
Run multiple SAP instances on the same Windows based HP Integrity server

Situation:   With the introduction of the Intel Itanium based HP Itegrity high-end 64-bit servers capable of running HP-UX, Linux and Windows clustering of these huge servers with up to 64 processorsockets and 128 cores, the desire for partitioning and virtualization came.
Customer took it for granded, that these huge servers could run SAP as good on Windows as they could on HP-UX. But that was not the case. SAP only supported a two node Windows installation.

Solution:   HP had Serviceguard for HP-UX to manage many instances of applications, such as SAP Application Servers, Replicated Enqueue and Oracle RAC. The thought of having something similar to Serviceguard Extension for SAP (for HP-UX and Linux) available for Windows started. The result of these thoughts was development of Competent Cluster Solution.

Result:   : HP Competent Cluster Solution ( HP CCS) makes it possible to run and easy control multiple instances of SAP, Replicated Enqueue and Oracle Real ApplicationCluster. Using HP CCS the potential of HP Integrity server's virtualization is fully utilized.

Read the Press Release (pdf, 290KB)
Read the Service Brief (pdf, 99KB)
Read the Data Sheet (pdf, 109KB)
Read the Customer Solutions Guide (pdf, 396KB)
Read Jürgen Thomas, Microsoft, blog, posted May 30,2007 SAP Multi Instance Clustering (pdf, 109KB)

Automotive tyre company, Germany, 2005
Pilot project: SAP HR system with Oracle RAC on SuSE Linux

Situation:   Customer wanted to investigate into clustering technology using Industry Standard Servere for higher availability and reduced cost for smaller system. Customer wanted to pilot SAP on Oracle RAC running 32-bit SuSE Linux.

Solution:   Customer agreed to be a pilot project for SuSE Linux using HP Serviceguard and Serviceguard extension for SAP (SGeSAP) to implement Oracle RAC and protect the SAP Enqueue Server in a Linux Cluster. SGeSAP were by my team ported from HP-UX to 32-bit Linux and implemented.

Result:   Customer got a high available SAP system where both the Oracle Database and the SAP Enqueue server were clustered.

GAS Natural, Spain, 2005
Pilot project: SAP SEM with Oracle RAC on Red Hat Linux to replace a HP-UX system

Situation:   The customer wanted to replace their aged HP 9000 HP-UX system running SAP SEM with a newer system giving better response time. Customer already had quite some Oracle RAC implantation for non SAP systems in production and wanted to evaluate SAP on Oracle RAC running Linux on industry standard severer.

Solution:   Sizing showed, that a SEM system consisting of seven Oracle RAC database severer and one SAP application server should be able to do the job. A setup using Oracle Cluster File System was installed for a pilot project.

Result:   : The initial sizing showed to be correct. The new system had two-three times better answering time and better price-performance than both aged HP-UX system as well as an expected HP-UX configuration on new HP Superdome technology.

Summit Strategies has release a released a report describing SAP customers use of Oracle RAC. The GAS Natural project is descibed in the report SAP Customers Depend on Oracle Real Application Clusters for High Availability and Scalability (pdf, 204KB)

Read the Oracle Success Story (pdf, 72KB)
Read the Red Hat Story (pdf, 131KB)


Lengthy restore time, 2005
Situation:   Customer experienced very long time to restore their SAP / Oracle database.
Due to the huge size of the database, the daily backup window was too small to do off-line backup. Customer did full off-line back during weekend and on-line backup daily. Due to the large database and the load on the system only half of the tables could be backed up during the daily backup.
Restore were time consuming either due to several days old off-line backup and large amount of log files has to be applied or inconsistent on-line backup with large amount of log files to be applied.

Solution:   Changed the system to do daily snapshot of stopped and thereby consistent database with only a max of less the 24 hours of log files to roll the database forward to point in time. The window, where the SAP system were not available were around 15 minutes (stopping the database, create the snapshot and restart database).

Result:   Restore from tape created a consistent database that eventually needed to be rolled forward a few hours to be up to time.

HP StorageWorks for mySAP Business Suite papers, 2004
Listed below are three papers, which were created in the storage team under my management. I personally wrote some of them in their first incarnation and provide text to other of them.

The concept of Rapid Backup for SAP was the reason for starting the the team. Below you find two pdf-files, the first is the Business Blueprint HP StorageWorks Rapid Backup for mySAP Business Suite for EVA (pdf, 395KB) describing the business reasons for implementing the solution.

The second document is the solution guide HP StorageWorks Rapid Backup for mySAP Business Suite on EVA for Oracle (pdf, 553KB) , describing the function of the solution.

Several other storage related papers available today in latest versions, contains text initially created by myself or in the team under my management, of which the following three can be mentioned:

HP StorageWorks Enterprise Virtual Array configuration guide for SAP Business Suite (pdf, 574KB)

HP StorageWorks XP Disk Array Configuration Guide for mySAP Business Suite (pdf, 897KB)

HP StorageWorks Modular Smart Array configuration guide for SAP Business Suite (pdf, 359KB)


ABB, Manheim, Germany, 2003
World Wide first SAP with Oracle RAC implementation

Situation:   ABB wanted to analyze the possibility to enhance availability of several of their SAP systems. As they already had some experience with Oracle RAC on non-SAP systems and heard about Oracle, SAP and Compaq's plan to get SAP on RAC certified, ABB volunteered to be the first pilot customer for such configuration.
An engineering task force from Oracle, SAP and Compaq was formed with me as the head from Compaq. Together with an additional Compaq engineer and engineers from SAP and Oracle numerous tests with customer data were done at the premises of ABB were done.

Result:   ABB were the world first SAP customer that went productive on SAP on Oracle RAC. Due to this successful pilot, certification of SAP on RAC was graded. Later additional SAP systems at ABB were enhanced with Oracle RAC.

Read the HP press release.
The configuration is described in Oracle newsletter No 12. - SAP News 2003 (pdf, 1.499KB)
At Deutsche ORACLE-Anwendergruppe e.V. ABB's Jürgen Wingerter has posted a presentation Running SAP R/3 on Oracle 9.2 Real Application Clusters at ABB) (pdf, 388KB)


Petsmart, US, 2000
Petsmart SAP Implementation Was Huge Hardware Challenge

Situation:   After implementing a new SAP Retail system on a new Tru64 system with EMC storage, the system was not able to handle the daily retail information in 24 hours.

Solution:   analyzed the system and pointed to the fact, that the storage system was not able to handle the large amount of I/O. Both the number of I/O channels as well as the storage subsystems I/O capacity should be increased. After a long period the customer chose to replace the EMC storage with Compaq storage sized and designed by me, the system performed extremely well.

Result:   regained customer satisfaction and provided a very powerful Sap Retail system to the customer.

Article from AllBusiness describing the challenge here.


Pioneer Food /Sasco, South Africa, 2000

The South African food manufactures, Pioneer Food Group implemented in 1999 - 2000 a Compaq ProLiant 8500 based SAP system. The system was running Windows NT and Oracle as the database system. The database server was connected to a Compaq StorageWorks storage system. After rolling out SAP to additional branches, the SAP system was believed to be the world largest SAP system on Windows NT with the world largest SAP Oracle database.

Situation:   Huge business growth due to acquisition. Customer was facing longer database response time due to monthly data growth of roughly 30GB. Customer needed help to design SAP infrastructure for large database growth and high availability / disaster tolerance. The customer was running SAP on Windows NT and with Oracle as the database.
Growing the Oracle database included often installation of additional disk spindles and creation of new LUNs that were concatenated under the OS.

Solution:   Analyzed the configuration and suggested change in layout of virtual storage as well as how newly installed storage capacity was introduced to Windows. I defined optimal storage layout and practices for future management. Result: as database size grew more spindles were installed resulting in stable response time. Actually response time got slightly better over time due to additionally spindles increasing the total amount of I/O of the disk system.
Together with the customer's hardware provider I helped designing a dual data center setup, where data were replicated via dark fibre from one data center to another.

Press release from Compaq/ HP here.


Deutsche Post AG, Darmstadt, Germany, 1999

The German mail company Deutsche Post AG (DPAG), had implemented the world largest SAP R/3 installation on Windows NT. Many different systems and instances were running. They all ran on Compaq Proliant with Compaq SCSI connected storage. For the larger systems the storage were external to the ProLiant connect via SCSI to one or more Compaq SmartArray disk controller. During the development of Compaq's first fibre attached storage, DPAG were contacted for being a pilot customer. DPAG saw the possibility to reduce the space occupied by storage, as more compact storage could be deployed due to thinner disk drives of larger capacity. During the period of testing DPAG started on a project on High Availability, where they ended up with the desire of having tow computer centers interconnected with dark fibre.

Situation:   DPAG's main data center were located close to Frankfurt Airport. Under certain weater conditions, large numbers of airplanes flew over the centre during take-off or landing. Due to business criticality, DPAG wanted to have their SAP business applications implemented with as high availability as possible with respect to cost of implementing such.
A concept of two computer centre (Ausfall Rechenzentrum) was developed where data from one centre should be replicated to the other centre.

Solution:   After Compaq's aqusition of Digital Equipment Corporation Compaq suddnely had a proven solution with Enterprise Modular Array.
After Compaq's acquisition of Digital Equipment Corporation Compaq suddenly had a proven solution with Enterprise Modular Array.
The EMA1200 were introduced and after testing an order of roughly 30TB storage were installed (in 1999 / 2000, 30TB was HUGE).

Success story (in German) "Da Geht Die POST Ab" (pdf, 1.09MB)


Qualification of StorageWorks Data Replication Manager for SAP, 1999

To be able to offer replication of storage for Deutsche Post AG and other customers, it was vital to ensure, that the replication of data would work under high SAP load. A way to prove was to have the StorageWorks Replication Manager running during a SAP R/3 high-load benchmar, which were being certified by SAP.
I initiated such brenchmark executed by iXOS in the data centre of DPAG in Darmstadt. The results were certified by SAP.

News release: Compaq Offers One of the World's Fastest SAP Disaster-Tolerance Storage Solution read here


Large LAN Manager installation, 1994

I was involved in one of the largest LAN installation in Denmark at this time. A large Engineering Consultancy company wanted to share printers and storage for their 1.100 - 1.100 employees having a PC. The company used large Prime UNIX machines in their CAD installation. They had initiated a transition from this UNIX based CAD application towards using AutoCAD on large PCs. I analyzed their requirements and wrote the proposal for implementing a LAN Manager based net, that had connection to the Prima server as well as multiple file and print servers. For the printers dedicated small network based printservers, that could accept print streams both from a LAN Manager servers well as TCP/IP stream directly from UNIX servers were proposed. Remote connections to around 10 remote offices were also offered.

To be able to route the information to and from the remote offices, LAN Manager on top of TCP/IP were proposed. At that time, Microsoft did not offer TCP/IP protocol for LAN Manager. We offered FTP Software's PC-Connect as the TCP/IP protocol stack. I wrote installation scripts that enabled the installer to chose either Microsoft NetBIOS/NETBEUI protocol or FTP Software's TCP/IP stack to be installed.


X.21 tester, 1986

Together with a friend, an x.21 tester was developed for the Danish Shell Oil Company. Shell was implementing communication from their gas stations to a central Tandem machine. The communication should be X.21. To be able to simulate non proper X.21 packaged, an X.21 package simulator was wanted. A system was constructed so all type of X.21 packages could be created and send to the central server. It was possible to send packaged, that wasn't a full package, simulating bit-errors, etc.


Build a 1mbps Ethernet, 1982

My final project at my B.Sc. studies at the Danish Engineering Academy, together with a follow student we designed and build prototype of a 1 mbps Ethernet. One of the prototype boards were build for my S-100 bus based Z80 microcomputer. I can't remember which bus the other was build for.

We wrote drivers for the CP/M operating system shoving, that we from one computer could access the (floppy) disk at the other computer. Amazing!


Importing/building my first microcomputer, 1980

As a hobby, in 1980 I assembled my first microcomputer. As parts ware not available in Denmark, I imported parts from the US and assembled it my self. The microcomputer was a S-100 based Zilog Z80 based computer. (The Z80 were compatible with Intel's 8080 CPU but were enhanced with additional instructions and registers. Additionally it was more advanced in its way of communicating with I/O equipment compared to Intel's CPU). I bought the CP/M operating system (if we can call it an operating system, it was more a BIOS with a bit more enhanced commands for handling I/O to disk (floppy disk), directly from Digital Research. Amazingly, I got the system up running.

I used this computer for my three last projects in my M.Sc. study in Computer Science. I primarily used Pascal MT+, MS-COBOL and MASM-80.

Read more about the S-100 and IEEE-696 Bus here.
Read about Digital Research and CP/M here.
Short information on Pascal/MT+ can be found here.