Yesterday, I presented a technology roadmap for I.S. Clinical Development (CCC) that will improve the infrastructure of their in-house “colossal” healthcare information system for high availability and disaster recovery. To everyone’s delight, there were no disruptive technologies or controversial implementation details to make this presentation last longer than the 10-minutes requested. Key points follow:

Implement Caché ECP

The “Enterprise Caché Protocol” feature for Caché applications and databases allow it to be deployed in a multi-tiered fashion.  To avoid as much application re-work as possible, it was necessary to provide advanced clustering features in Red Hat Enterprise Linux, namely:

  • Linux Virtual Server for its IPVS capabilities of managing and distributing the variety of client connections into CCC, such as classic telnet, web, ftp, etc.;
  • Global File System for natively shared SAN access to common filesystems; and
  • Red Hat’s Cluster Suite that provides the above tools with an associated tools and toolkits to provide easier Resource Group management and monitoring for fail-over: IP addresses, filesystems, scripts, etc.

Upgrade Caché 5.2 to 2008.2

In addition to avoid lagging behind in InterSystems product releases for support, there are significant security and fail-over to shadow features with further enhancements to ECP to make this a worthwhile release to upgrade to.

Move from Symmetrix to Clariion storage

In addition to avoiding costs to a storage frame that is nearing its end-of-life, Clariion will provide us with MirrorView / Sync features to enable SAN-based replication between their frames at our production and disaster recovery datacenters.  Couple that replication with their “snap” feature, and we can monitor, inspect, and even do testing on the disaster recovery version, without disrupting the concurrent replication of production systems.

Rack to Blade server migration

This move will have significant impact on our future operations.  The IBM BladeCenter provides for high availability and advanced management functions of every component within its compact 9U chassis.  It truly provides for our growing needs of lights-out hardware management and operations.

We recently configured its Open Fabric Manager that re-programs every ethernet and fibre channel card to fixed MAC and WWPN addresses.  This allows for easy replacement and upgrade of I/O components without the need to re-zone SAN disks or reconfigure the networking stack on the host.