Study

The scale-out option offered RI-Solution major
advantages. “Both concepts would have worked from
a technical perspective”, admits Hanns-Gunter Weber,
Head of IT Infrastructure at RI-Solution. “However, in
terms of the service costs, the HP bid differentiated
itself significantly from the alternative.” With a scale-up
approach, the company would have had to invest
significantly more in high-value services, as an outage
within the central host system would have had a far
greater impact than an outage within the decentralised
scale-out architecture. The user would have been reliant
upon an expensive service, which would have to react
extremely quickly to outages and would have been
significantly more costly.
By contrast, because the load is spread over several
smaller systems in the scale-out approach, the risk is
more widely distributed. Even in the case of widespread
hardware failure, RI-Solution would be able to insert
several blade servers into a functioning enclosure to
restart business critical production systems. As it is not
entirely dependent on an immediate resolution of the
hardware fault, this lowers the service costs.
The scale-out concept also offers benefits in terms of
unplanned growth. In the past, RI-Solution came up
against the limits of the very large, indivisible systems.
This made large investments in major new systems
necessary. With the scale-out approach, it is possible
to expand the infrastructure on a piecemeal basis. In
addition, RI-Solution can take a more flexible approach
to servicing a high number of blades, which can be
swapped out among several enclosures. When a host
system needs to be shut down for maintenance, all
applications are halted.
“We had the choice between two fundamentally
different architectural concepts,” Weber concludes.
“A comparison of the overall costs demonstrated that
HP’s offer was economically superior. It convinced us
to such an extent that we took the risk of changing
our service provider for our SAP system configuration.
We minimised this risk by awarding the whole SAP
configuration migration to HP as the general contractor.”
SAP and non-SAP systems migrated to
blades
The SAP database and central server are now running
on 32 Integrity blades. In parallel, non-SAP systems were
migrated from 196 older HP servers, most of them rack
servers, onto 131 blades from the newest HP ProLiant
generation. Additional SAP application servers were
integrated into the common blade configuration of
Primary Hardware
• 2 x HP Superdome 2
• HP Integrity BL860c Blade
• HP Integrity BL870c Blade
• HP ProLiant BL495c Blade
• 6 x HP Storage 8000 Enterprise Virtual Array (EVA)
• 2 x HP ESL Tape Libraries
• 2 x HP VTL9000 Virtual Tape Library
Primary Software
• HP Systems Insight Manager
• HP Serviceguard Solutions for HP-UX
• HP Virtual Connect Enterprise Manager
• HP Data Protector Software
• DB2
• SGeSAP with DB2
HP Services
• HP Critical Services
Customer solution at a glance:
ProLiant/Linux. The BayWa SAP retail ERP system, one of
the largest SAP-DB2 retail systems worldwide in terms of
the load demands, was installed on two high-availability,
high-end machines from the HP Superdome 2 series,
as was the Business Warehouse system. Due to the
extremely large workloads and the immense amount of
data to be processed, the obvious choice in many cases
was the manufacturer’s flagship Unix system.
The project team carried out the migration during the
holiday period because only 40 per cent of the usual
load is placed on the BayWa system at that time.
Standardised management for Unix
and the x86 world
After working through several glitches during the
implementation stage, RI-Solution achieved all of the
objectives of the project. Overall, it was possible to
standardise the operational and service processes as
planned. Among other things, it is now possible to
monitor the entire hardware infrastructure, both the SAP
and the non-SAP systems, via a single console.
In the meantime, the quality improvement far
exceeds monitoring alone. “Due to greater levels of
standardisation, the currency of the system, in terms
of firmware, patches and the like, has increased
significantly,” says Weber, drawing attention to another
key benefit. “As a result we are significantly closer to
achieving our objective of ensuring the highest possible
availability through proactive management, quicker
fault diagnosis and faster recovery following a fault.”
In summary, the system management now achieves a
higher quality at every level
3