September 29, 2011, 7:42 PM — A healthcare software provider has upgraded its Gigabit Ethernet network to improve service to customers and reduce errors during backup.
More on healthcare technology: High-tech healthcare technology gone wild
HealthMEDX , a Missouri-based provider of software for long-term care, home health and rehabilitation organizations, didn't increase the speed of its backbone. It merely swapped out switches from Cisco to Brocade to achieve wire-rate performance, which eliminated the backup errors and helped save the company's reputation with its customers.
"We were facing port errors every single night," says Chris Bingham, vice president of infrastructure at HealthMEDX. "We were facing bandwidth restrictions or available capacity restrictions. We had to meet maintenance and application delivery requirements and development requirements. We needed a faster network."
HealthMEDX replaced 18 to 20 Cisco Catalyst switches, anchored by two oversubscribed Catalyst 4507s in the backbone, with an equal number of Brocade FCX Series and Brocade FastIron WS Series Power over Ethernet (PoE) LAN switches. The company also installed Brocade-labeled wireless LAN access points - which are OEMed from Motorola.
The switches interconnect 100+ physical servers and 200 to 300 virtual servers, Bingham says.
HealthMEDX also evaluated HP Networking switches when undertaking the upgrade. HP is HealthMEDX's server vendor.
The Brocade system came in 35% to 40% lower in price than the Cisco bid, Bingham says, and the company expects a return on investment in 14 months. HealthMEDX has been using Brocade SANs since 2009 and the company says it completed the network transition in less than six hours without any downtime or disruption to users.
"From a financial standpoint, it was one of the best decisions we ever made," Bingham says.
HealthMEDX had a number of backup activities that occurred in parallel over the network and those jobs were failing, Bingham says. Switch ports would produce thousands of port errors -- an uplink port with 3,000+ receive errors, and 700 or 800 discard errors per port channel group.