Cut Through the Hardsell...

When we conceived of the Hospital Information System (HIS) for Fortis Heart
Institute, a super-specialty cardiac care center, we were, for the first time in
India, aiming to set up a near paperless and ‘filmless’ hospital. The target
was to have every department and every patient record (including images and
data) electronically linked and archived, to be accessed by administrators and
doctors from anywhere in the campus or over the net.

Choosing and implementing the right server configuration was of utmost
importance and this posed quite a few challenges. The first was to have near
100% uptime, as all the medical records of the hospital would be maintained
digitally, and even the shortest downtime of the database would have serious
consequences. So, we had to make sure that there was a buffer for fault
tolerance and redundancy in data.

The application software was NT based, hence limiting our options. Then there
was the problem of the geographical location, as the Institute was in Mohali and
most of the major vendors did not have a large set up there. In fact, some of
the vendors did not even have direct representation, which was a real problem in
terms of initial requirement understanding, supply, installation and finally,
support.

To meet the near 100% uptime, we thrashed our the issue and finally went in
for a clustering solution. We built in redundancy on each and every point –two
identical servers working on heartbeat between them, operating systems mirrored
within each of the servers, redundancy power supplies and a network card in the
servers and an external storage unit. Redundant controller with Raid 0 and Raid
1 has been implemented on external storage for database redundancy. This
configuration gave us the first level of redundancy. Repeated discussions with
the vendors’ marketing and technical teams were held to reach a comfort level
between the operating teams on both sides with respect to the concept and
configuration.

We also took the application software to the vendors’ research and
development centers to get the clustering/application checked out and there were
learnings for all of us in the process. This helped us to ensure that the
vendors understood and appreciated the criticality of our requirement. Then,
when the concept and configuration were in place, came the issue of 24x7 direct
support from the principals.

The principals wanted the reseller and the application service provider to be
part of the contract. We were clear that any hop in between would result in a
time lag, which we could not afford, hence we insisted on working directly with
them. We also sat down and worked out a separate service and support contract,
as part of the main order. The learnings from this experience would probably be
valid in choosing and implementing a high-end server configuration in most
organizations.

Our learnings were three-fold:

• Always go to the core technical team of the vendor,
  instead of relying on the company’s marketing team. This saves time and
  effort;

• Insist that the concept (requirements, application
  response and uptime) are clearly noted on paper, apart from the bills and
  other relevant paperwork. This would put in clear perspective the vendor
  understanding as well as the material requirement; and

• Always deal directly with the Principal and not his
  representatives while negotiating. This will avoid future confusion and cost
  or time escalation.

The author is head, Information Technology, Fortis Heart
Institute