The BMA cures its storage growing pains

The British Medical Association acts as both a trade union and professional association dedicated to protecting individual members and the collective interests of doctors. Founded in 1832, the British Medical Association is a UK organization formed “to promote the medical and allied sciences, and to maintain the honour and interests of the medical profession”.

The BMA is a voluntary and subscribed association. It currently has around 144,000 members in all branches of medicine across the UK, membership accounts for over two-thirds of practicing UK doctors. The BMA has a range of representative and scientific committees. It has sole negotiating rights with the UK government for all UK registered doctors. The BMJ Group, a wholly owned company of the organization, is a world-class producer of medical journals, issuing 26 specialist monthly publications in addition to the weekly British Medical Journal. It also delivers innovative point-of-care medical information for clinicians worldwide.

As an information organization, the BMA needs to store increasing volumes of data on its members, policies, science and administration. The organization runs almost 1,000 desktops in its London HQ, three national offices and five regional centers; 140 of these are home workers and there are 25 overseas employees. Ensuring good performance and availability for growing volumes of user data was already becoming problematic, because the existing Storage Area Network (SAN) was reliant on end-of-life hardware.

The BMA was facing an even greater challenge around the storage requirements of its publishing division. As a publisher, the BMA needs to store and manipulate large volumes of data to support the on-time dissemination of medical information to a worldwide audience. The organisation is continually adding new online distribution channels and branching out into content-rich web delivery including podcasts and high-definition video streams.

Martin Kelmanson, Head of ICTS, says: “We were seeing significant growth in storage requirements, and the rate of growth was itself increasing. The existing SAN technology was a year beyond the end of its planned life, and we had growing concerns about its long-term reliability and cost of maintenance. Equally, it was clear that the existing environment would not support our plans to improve availability, performance and ensure effective business continuity.”

Diagnosing the issues
The increased growth rate for data at the BMA was putting pressure on the performance and availability of the organisation's existing disk storage resources. Equally, the BMA identified the need to unify its divisional storage planning into a single strategic data management policy that could better tackle the issues of risk management and business continuity.

In terms of the physical infrastructure, the BMA had a single SAN fabric split across two rooms to enable planned maintenance without the loss of availability. In total, the environment had 23 TB of usable disk space, but less than 1 TB remained free for use.

“The existing SAN was in year six of a planned five-year life, and we could not expand it any further,” says Frank Edwards, Infrastructure Services Manager. “Trending showed us that we had enough remaining capacity to accommodate just five or six months' growth, at which point we would hit a brick wall.”

Finding a cure
The BMA worked with IBM to deploy two IBM XIV Storage Systems, with 43 TB usable capacity on each. The XIV systems are synchronously mirrored at volume level to provide added resilience to specific key systems, and the new SAN fabric offers 4 Gb/s connections to the BMA's 198 servers. These run a variety of business-critical systems, including core human resources, finance and membership applications, Lotus® Domino® for messaging, BusinessObjects for reporting, and several Oracle and Microsoft SQL Server databases. The server landscape is 40 percent virtualized, with some 85 servers running as virtual machines on VMware ESX Server.

“The XIV solution met all of our business and technical requirements, and we had read a number of positive endorsements of the technology from other users,” says Frank Edwards. “It was clear that the XIV systems would offer ample scalability, and that they would both reduce our risk and improve the efficiency of storage services to users. The deployment was fast and smooth—IBM helped us to complete the work over the course of five weekends, on time and within budget. They also ran a series of workshops to transfer knowledge and best practices, and we very quickly got up to speed with the XIV technology.”

The rapid adoption and ease of use offered by the XIV solution was a significant advantage for the BMA, which has only a small IT infrastructure team and operational IT budget. The new mirrored XIV solution removes all administration around RAID setup and maintenance, giving the BMA extremely high levels of protection for data with zero impact on staff resources. It also delivers ultra-high I/O performance, even as data volumes grow.

“We are approaching 90 percent allocation on the XIV systems, and there has been zero decrease in performance against our baseline 110 GB of data,” says Frank Edwards. “The XIV solution offers stunning performance thanks to its massive parallelization of read/write operations.”

Performance and resilience
The introduction of the XIV systems enabled the BMA to consolidate and simplify its storage environment, replacing multiple disk arrays with just two XIV systems.

“The performance gains we've seen with the XIV systems are amazing,” says Frank Edwards. “For example, cloning one particular system used to take over 30 minutes, and it now takes less than 15 minutes. Both administrators and end-users have seen benefits from the significant improvements in performance for our Oracle production databases: some tasks complete in half the time when compared with the baseline performance. The XIV systems provide extremely high performance for our business-critical systems even as the volume of data continues to rise.”

“We have also seen the XIV's ability to cope with the loss of system components without degrading performance,” he adds. “Users were unaware that we had lost a module (12 disks), as there was not a blip on any of our systems when it went offline. Our Helpdesk, which would previously have lit up when we rebuilt even a single disk on the old system, was eerily quiet during the rebuild of all 12.”