Liverpool Heart and Chest NHS Hospital Trust has ended its ‘weekly’ data centre outages with a private cloud featuring EMC storage, offering the reliability needed to support the roll-out of a new electronic patient record system.
As one of the largest specialist hospitals in the UK, uptime for the trust's digital services is important to maintaining a high level of patient care. When it planned to roll out a new EPR system, it was decided that its legacy HP EVA SAN array did not have the capacity or reliability to support its primary application, said deputy head of IT, Matt Connor.
"We knew we needed to put in an electronic patient record system, but at that time there is no away we could have delivered an EPR system on that infrastructure. The log-on times were horrendous for a start, and there were inconsistencies from a user experience," he told ComputerworldUK.
“We would also have an outage a week prior to the IT improvements, with some systems that were failing or would need corrective action or a restart. There is no doubt it would affect patient services in some way."
He added: “Now in conjunction with the EMC technologies and all the other IT improvements we have made with new desktops, monitoring systems and so on, our outages are very few and far between. We have not had an outage for a long time and we are having no performance issues.”
In January 2013, the hospital went live with the first of two VNX 5500 storage arrays, each with 60TB and small amount of flash storage. The hospital went live with its EPR system the following June, before adding another array at a backup site later that year, in October. The storage investment cost more than £800,000, including integration costs through a service provider.
The implementation was part of a private cloud set-up, enabling replication across to a secondary facility for the first time, with a disaster recovery facility three mies away.
“We have been able to failover our EPR system in three minutes to our second data centre,” said Connor.
To support its various clinical and other applications, the trust also uses HP blade servers, 70 percent of which are virtualised with VMware vSphere. A physical HP ML530 database server is also used, running on Windows Server 2008 and Microsoft SQL 2012. It is also using Microsoft System Center Operations Manager (SCOM) for monitoring purposes.
HP 3Par lacks 'flexibilty'
Before deciding on the VNX arrays, the trust looked at other vendors, but found that they lacked the functionality of the EMC system.
“We did look at HP 3PAR technologies, and we looked at NetApp. They are good products, but for us, the EMC technologies met our needs best, especially as a supportive infrastructure of the EPR system,” he said.
“The storage virtualisaiton piece was big differentiator at the time [over 3PAR]. The HP systems weren't as flexible in terms of the virtualisation of the storage that we were after.
“The ability to use flash and tiered storage within the same array was something that was quite key to us. We are not a big site, and we can't afford to have multiple racks of storage - we wanted something that was quite compact with all of the relevant tiered storage in it, and the EMC system certainly met that."
The trust is also using a number of other EMC products including Atmos for cloud archiving, and Avamar for backups.
“With Avamar, not only does our backup window fit within the 12 hour night period before the 6am shift kicks in, our retention has gone from two weeks to six months, so we are able to apply a much lengthier period of retention on our backups.”
Having successfully launched its EPR system, the trust is targeting the roll out of virtual desktop infrastructure for its 1,200 staff, with a requirement for around 1,000 computers, having completed a pilot project.
According to Connor, the VDI storage is likely to be separate from its core VNX systems, and may run on VMware’s software defined storage platform vSAN.
“We are quite keen to understand what the VMware vSAN will offer in terms of distributed server storage,” he said.
“The VNX will support it, but we choose not to do so because it is a different workload and we want to segregate out VDI infrastructure off our main storage,” he said.
“We don't want to compromise our infrastructure given that the EMC systems are supporting our key critical systems, so mixing that workload doesn't make a lot of sense for us.”