System down? No problem: KVM-over-IP remote-access solution brings network back up quickly - Remote Access

Communications News, July, 2003

Hardware can fail. When a recent hardware failure had systems engineers at Washington, D.C., government facilities scrambling between locations miles apart, IT managers knew the time had come to find a better, more efficient way to manage and maintain systems across the district. They found their solution in digital KVM over IP.

For the Washington government, security, disaster recovery, redundancy and fault tolerance need to function at the highest levels and remain unencumbered by budget constraints. To ensure successful operation of its distributed data centers, the District of Columbia (DC) government brought in the expertise of EastBanc Technologies.

DC's systems are maintained in multiple locations by design. EastBanc worked with the DC government to replicate the data centers for the part of its network that is maintained for and used by the public, referred to as the "demilitarized zone," and EastBanc continues to maintain it. The replicated servers are kept miles apart so that if a power outage, system crash or other disaster damages one server, the others can pick up the slack. The EastBanc team that manages and maintains these servers was in yet another building, away from the demilitarized zone.

The constantly changing and expanding nature of the DC government's Web application infrastructure made maintaining the systems at multiple sites an almost overwhelming task. Finding an effective means for remote access to the servers and networking equipment in DC's multiple data centers became a priority. The district determined that it needed a scalable, enterprise-level, remote-access solution, and began the evaluation process.

SYSTEM ACCESS WAS DIFFICULT

"Our (number of) servers began to grow rapidly, not just one by one, but by the tens," explains Oleg Slivin, EastBanc Technologies' systems engineer. "With many servers in place, there was limited space in the racks. The solution that came with the rack was a standard local KVM switch. After a hardware failure, you couldn't even touch the box. You had to go over and physically work on it. It was then we realized that we would like to have console access to the server over the Internet. This way, if anything did go wrong, having remote access would allow us to get to the server console quickly, no matter what.

"There are some very sensitive services that are very crucial parts of the system, such as the Department of Motor Vehicles (DMV), that allow customers to make online payments, renew their tags or complete their vehicle registration," Slivin adds. "If the DC government portal goes down and stays down for a few hours, it's unacceptable. Having remote KVM access significantly lowers the time it takes to bring the systems back up."

EastBanc investigated three different vendors offering similar technologies, bringing the products into the lab and running them through the paces.

"We did a lot of preliminary work before we selected Avocent," says Slivin. "We wanted proven technology. Avocent had an advantage from the beginning--all the Dell racks are equipped with KVM switches from Avocent. The difference was that they were local versions, without the IP connectivity.

"The Avocent system takes extremely low effort to make it work," Slivin adds. "We put it in the rack, powered it up and configured the IP address on the KVM system. It took a half-hour to get it up and running and have control of the rack over IP from my desktop."

Slivin's team is managing approximately 60 servers via Avocent DSR2161 units using PS/2 converters. Since other teams at DC government use Avocent solutions to manage their server farms, total management is several hundred servers via Avocent equipment. In addition, DC government has several CPS1610 terminal servers to remotely manage network equipment via local console ports.

"The solution helps us reduce costs, improve uptime and increase performance," says Slivin. "We can now have remote access to all the servers in data centers from a single management station."

REMOTE REBOOTING NOW POSSIBLE

Before using the system, a hardware crash required the systems engineers to pack up their equipment, run over to the data center miles away and then physically find the server to update it. Now, Slivin can log into the KVM switch remotely, and even though he may be several miles away, reboot the machine as if he was physically pulling the plug and putting it back in.

"There's a clear time savings between sitting at your desk vs. getting up and running to a different location," he offers. "Servers don't go down that often, but they do go down. This solution has dramatically helped our uptime percentage and has greatly helped us work toward no downtime."

When one of the domain controllers at the remote facility had to be restarted after recycling the power and did not respond to network pings, Slivin was able to fully realize the power of KVM over IP.

"We knew for sure that the domain controller was down, so I connected to the server console using the KVM over IP switch to see what was going on, "he explains. "The diagnostic screen said that it was a hard disk failure and indicated that some settings were reset in the system BIOS. Instead of a disk array, we had three independent disks, which is a pretty scary picture. Using the KVM over IP, I was able to get into the server BIOS and change the settings back. I rebooted the server and it came right up. It was minutes instead of hours.