As documentation from doctors and nurses continued to blossom at the University Hospital of Leipzig, the hospital's IT department knew it had to protect its applications for the long term.
The Saxony, Germany-based hospital collects a steady stream of documentation to treat and bill patients in its 38 clinics, polyclinics and departments. While the hospital had a second data centre to replicate data from its critical applications -- SAP Human Resources and SAP Business Warehouse -- this wasn't reliable in the short term and wasn't a long-term solution to store years of data. The hardware that hosted these applications was at the end of its lifecycle and the facility had suffered two previous data centre outages, according to Daniel Pfuhl, the head of department for systems management.
An outage can cost the hospital 600,000 euros a day, said Pfuhl, if members of its 3,000-person staff cannot access necessary data. In the long run, the IT department knew that, by law, it had to store patients' records for a minimum of 30 years.
"The sheer amount of money the hospital could lose from an outage ... was our main argument for the investment in a new DR strategy," Pfuhl recalls.
Enter the hospital's IT project, "Red.IT" (or Redundant IT Infrastructure).
Data centre rebuild eases IT pain
The solution to the hospital's IT pain was to build two new data centres, both based within the University Hospital's grounds, with a combination of new Intel-powered hardware and virtualisation software from VMware. By doing so, the hospital can now perform failover and failback tests twice a year.
The sheer amount of money the hospital could lose from an outage ... was our main argument for ... a new DR strategy.
Daniel Pfuhl, the head of systems management, University Hospital of Leipzig,
Red.IT has virtualised 95% of the hospital applications, and the availability of these applications has improved, according to Pfuhl. Other virtualised applications used by the clinic include Exchange 2007, SQL Server 2000 and 2005, and Citrix Presentation Server 4.5. The main products used to build this new infrastructure were vSphere 4.1, SRM 4.1 and EMC Networker 7.6.
A DR strategy built on virtualisation
The hospital wanted encrypted failover based on virtualisation and for every mission-critical application to be protected by just one transparent failover mechanism, which can be tested twice a year. "We now have one failover configuration, which is safer and simpler to manage," Pfuhl said.
With VMware virtualisation, the Red.IT project has virtualised 95% of hospital applications, and through the use of VMware Site Recovery Manager (SRM) the availability of these applications has improved, according to Pfuhl. The hospital has 50 physical servers across both locations and a total of 450 virtual servers. Previously it had to run and manage 150 physical servers.
Pfuhl said the main goal of the project was to standardise on VMware based on a previous proof of concept in 2004.
"From the start there was a clear understanding that this project would focus on virtualisation and VMware, because we previously underwent a pilot project with ESX and vCenter. We felt this project was successful with some good benefits, so we made virtualisation a major goal," he said.
Problems with cooling and space
In addition to improving its data protection strategy and application availability, the hospital had to address data centre cooling and space problems. The old data centre's temperature had increased, said Pfhul, and that there was no way to monitor it remotely.
With virtualisation the Red.IT project has virtualised 95% of hospital applications.
"You had to go in and feel for yourself or check a basic thermometer," he said.
The temperature was, on average, 25 to 26 degrees Celsius in the hospital's old data centre, according to Pfhul. But, at its highest temperature, the data centre could reach more than 30 degrees Celsius. Today, however, at rack level the temperature is 17 to 18 degrees Celsius.
The organisation's primary and secondary data centres use Knürr rack systems with water-cooling lanes on the front. The IT department also uses its own system of fans at both sites. The temperature can now be monitored remotely.
Pfhul said that power has also risen, so the major investment in physical severs was also an effort to reduce power consumption. The consolidation of its physical servers meant more space. The secondary data centre site has 60% more capacity than its primary site.
Dr. Gert Funkat, --who, along with Pfuhl was one of the Red.IT project leaders -- said the SEVIS application that doctors use for cardiology, means large images need to be stored.
"The new sites are bigger, but we currently only use a third of the space," Funkat said. "The extra space will be used for archiving and storage technologies, as doctors and nurses are using more images and videos for treatment."
The team's plan is to fill this extra space over the next five to 15 years, knowing they have to legally archive the hospital records for a minimum of 30 years.
From start to finish the project took three years to complete. On Sept. 9, 2010, the IT team verified that failover and failback with SRM worked for 115 production VMs. Given a high degree of automation, today the hospital can now maintain two data centre locations with the same staff that three years ago ran just one data centre.
For more VMworld Europe 2010 conference coverage, click here.
Kayleigh Bateman is the Site Editor of SearchVirtualDataCentre.co.UK.
Postrcript: The specs for Red.IT
- For its Red.IT project, the University Hospital of Leipzig used a combination of Intel-powered hardware and VMware vSphere 4.1. The hospital also enlisted VMware Site Recovery Manager (SRM) 4.1, and EMC Networker 7.6 for its virtual infrastructure.
- The clinic also uses virtualised Microsoft Exchange 2007, Windows SQL Server 2000 and 2005, and Citrix Presentation Server 4.5.
- A fewer number of servers now support 5,000 physical desktops and 20 virtual desktops.
- Opting to use Fujitsu RX600S4 with Intel Xeon 7350/7460 and RX300S5, the hospital uses VMware ESX hosts (21 in total).
- These servers are attached to EMC CLARiiON CX4-960, CX4-480 SANs and Brocade Fibre Channel switches, in addition to two DataDomain DD880's for storage.
- The hsopital also uses VMware vCenter, VMware vMotion, VMware Distributed Resource Scheduler (DRS) and VMware High Availability (HA).
- "Red.IT" deployed its environment on Windows 2008, Windows 2003, Windows 2000, Windows XP, Linux SLES, debian, Ubuntu and Open SUSE.