Cisco UCS Implementation In 30 Days? VAR Makes It Happen

World Wide Technology, a solution provider and early adopter of Cisco's UCS technology, put its experience to good use in a deployment to help a radiology services provider store and migrate about 10,000 radiological images a day.

Working with Cisco, VMware, and NetApp, St. Louis-based World Wide Technology built a complete virtual data center infrastructure in only 30 days, making it possible for NightHawk Radiology Services to not only manage its existing customer base, but also prepare for an expansion of its services offerings to a completely new set of customers.

That infrastructure was built around Cisco's Unified Computing System, or UCS, which combines networking, blade servers, storage, core switching, routing, security, and voice over IP (VoIP) into a single architecture.

NightHawk Radiology Services is a Scottsdale, Ariz.-based provider of radiology interpretation services to about 780 radiology providers serving about 1,600 hospitals around the country.

The company provides over 3 million radiological studies each year. Each study consists of a hospital or imaging center sending its radiological images to NightHawk, which then forwards them to one of its team of radiologists who interpret the results for the clients.

Each image averages between 30 MBs and 50 MBs in size, meaning that NightHawk needs to be able to handle the receipt of 500 GBs of data each day, and forward about as much data to radiologists around the world to do the interpretations, said Ken Brande, vice president of IT for the company.

Because NightHawk stores between nine and 12 months worth of data, it requires about 120 TBs of compressed capacity to handle the load, Brande said.

Unfortunately for NightHawk, the company built its infrastructure too quickly, with little thought paid to a common architecture, Brande said.

"We deployed technology as needed," he said. "Our customers are mainly domestic, but our radiologists follow the Sun, and include members in places like Sydney and Zurich. So we had a wide range of server rooms, different power requirements, and different ways of error handling. That lead to service problems. In a remote office, using a toaster might crash a server."

NightHawk had a colocation facility in Chicago which housed most of its old technology, but had no room to expand. As a result, many of its servers were deployed in remote locations, Brande said. So the company opened a new facility in Phoenix and decided to implement a virtualized environment for the flexibility and the ability to control costs, he said.

Next: Cisco Intros UCS And VAR To NightHawk

To get such an environment, NightHawk turned to its networking provider, Cisco, for ideas. Cisco introduced the company to its UCS architecture. UCS competed against the company's then-existing architecture, which was built around Dell blade servers, and got the nod from NightHawk in May or June of 2009, Brande said.

Cisco also introduced NightHawk to World Wide Technology, which took over the lead on the project to configure, sell, and deploy UCS in NightHawk's Phoenix facility.

Cisco has brought WWT in on several UCS deals in part because WWT was Cisco's first UCS-certified partner, said David Harrison, director of WWT's Cisco data center practice.

"The deal with NightHawk came from us being tied to the hip to Cisco," Harrison said. "Also, our sales guy had a good relationship with Cisco and NightHawk."

Harrison said his company was one of the first to embrace UCS, and for that reason the design and implementation was pretty straight-forward.

In fact, Harrison said, the only real issue was the 30-day time frame in which the implementation had to be done. "[NightHawk's] time frame was very aggressive," he said. "They were very clear. They had less than a month. It all came down to logistical orchestration, and went very well."

The reason WWT could handle that orchestration stems from the way it works with customers, Harrison said.

WWT has five teams focusing on different specialties, including facilities and infrastructure; storage, backup, and recovery; networking; compute nodes; and virtualization and cloud computing. While each team is specialized, they also work closely with each other, Harrison said.

"We have a very structured planning process," he said. "Data centers are more than just networks. There are a lot of pieces that need to work together. We work with these teams to offer a certified solution.

Next: The Solution

NightHawk started at the top with two Cisco Nexus 6120 XP fabric interconnect appliances, which are the heart of a UCS implementation. Connected to those devices are four UCS blade server chassis in which a total of 18 half-width Cisco B200 blade servers are connected.

Those servers are uplinked to a pair of Cisco Nexus 5000 switches, which connect server, network, storage, and facilities assets together. They are connected via a few Cisco Nexus 2000 switches to non-UCS infrastructures.

The implementation also included a pair of Cisco Catalyst 4900M switches Gbit and 10-Gbit Ethernet switches, as well as a couple of Cisco ASA 5540 Adaptive Security Appliances for handling VPN and security.

In addition to the Cisco equipment, WWT also deployed virtualization technology from VMware along with storage appliances from NetApp. Tying the virtual servers to each other, the physical servers, and storage were multiple Nexus 1000 virtual switches configured by WWT.

In addition to managing NightHawk's radiological studies, the UCS infrastructure also runs the company's back office applications including Exchange, Active Directory, Oracle databases, and internal Web solutions, said Chris Smith, manager of NightHawk's data center infrastructure.

Looking forward, NightHawk is planning to take advantage of its new UCS infrastructure to expand its business to provide a quality assurance service for radiological image interpretations done by third-party interpreters, Brande said.

The company already has a quality assurance process for its own interpretations, and can offer it as a service to others, Brande said. "This comes from the agility we now have to bring up the assets we need for a new service quickly," he said.