Reducing costs and getting more out of existing technology has become a mantra for IT professionals, especially those with mature networks such as frame relay. Companies are deploying a variety of frame relay management tools that not only help get a handle on costs, but also improve performance.
Reducing costs and getting more out of existing technology has become a mantra for IT professionals, especially those with mature networks such as frame relay.
Companies are deploying a variety of frame relay management tools that not only help get a handle on costs, but also improve performance.
Although United Parcel Service (UPS), Staples and the South Carolina Army National Guard operate different types of frame relay networks, all have benefited from using products from NetScout, SolarWinds.Net and Visual Networks, respectively.
UPS deployed NetScout's nGenius Performance Manager to get a better, more detailed look at its network. One goal was to pinpoint which applications were driving bandwidth consumption, says Peter Gunn, project manager at the shipping company in Mahwah, N.J.
Although Gunn says it's difficult to quantify total savings since deploying the product, his intent is "to slow the pace of increasing costs and keep network costs at a reasonable and manageable level. I believe we've been able to do that."
UPS operates 51 frame relay networks that connect 2,500 sites worldwide. The company works with a handful of service providers, including AT&T, Equant and Sprint, to support that architecture. Although keeping a close eye on the network always has been important to Gunn and his team, UPS' older management tools didn't meet all the company's needs, he says. UPS had used Concord Communications' Network Health management tools from just about the time it set up its frame network in the mid-1990s.
"Previously we had a private network where we owned the circuits and could see into the network. With frame relay a public network, we didn't have that type of visibility," Gunn says. Two years ago, UPS decided it needed a more granular view than the Concord product could provide, he says.
The company was looking to identify how much bandwidth specific applications consumed and at what rate its bandwidth consumption was growing.
After Gunn deployed the NetScout system, he gathered this information and took corrective action that has let UPS avoid bandwidth increases at certain sites. "Cost avoidance is as important as cost reduction," Gunn says.
"Our objective was to map our business applications regardless of protocol. We wanted to see it in terms of things that are well known and understood by the business community," Gunn says. NetScout provides that level of detail, where e-mail is defined as such and not simply as POP3 traffic, he says.
Instead of throwing more bandwidth into the network, UPS deployed Check Point's FloodGate-1 policy management tool. This lets Gunn keep applications such as e-mail, by far the leading bandwidth hog on the network, in check by ensuring it never consumes all bandwidth at any given site. Gunn says some departments have taken further action by reducing the size of individual mailboxes and restricting the size of attachments.
Like many network managers, Gunn has joined together a handful of products to meet his monitoring needs.
The reason some departments have taken their own actions is because Gunn also uses the tool to calculate departmental chargebacks for application usage. Before deploying NetScout, UPS estimated application usage for each department. "Some departments were very happy when we changed our chargeback methods. Others were not as happy," he says.
UPS recently started using additional modules included in the NetScout product to improve network visibility from specific sites on the network.
"There is a newspaper module that allows us to publish [network] activity the following day for any given location in our environment," Gunn says. "Our field organization is quite extensive with 4,000 folks worldwide."
This feature lets those users see unusual or abnormal network behavior on a specific Web page for their location, which might not affect performance but is worth getting a handle on before it does.
Gunn chose the NetScout product after much research and in-house testing. And while UPS has been happy with the product, Gunn says that with any project there are issues.
"Right out of the gate, we had scalability issues," Gunn says. "But we worked very closely [with the vendor] and then migrated to the next-generation product, which pretty much established a good stable environment."
Gunn recommends that when shopping for a product users should consider the specific issues in their network environment. There were NetScout reference customers that appeared to be managing significantly larger environments than UPS, so Gunn assumed that scalability would not be an issue. "And it turned out to be our biggest issue," he says.
Different challenge at Staples
Like UPS, Staples is runs a large frame relay network, but capacity planning was a bigger issue for the chain of office supply stores. The Framingham, Mass., company deployed SolarWinds.Net's Orion management tool to get a better handle on utilization across its 1,200-node frame relay network.
Staples used HP OpenView, which provided the status of each circuit and alerts when a problem occurred, but the company needed more, says Sally Jo Bernard, director of network operations.
"We needed a tool that would give us capacity information so we can upgrade and downsize circuits," she says.
When Staples deployed the product about 18 months ago, Bernard and her team monitored capacity on all frame relay connections at 256K bit/sec or above. They determined that many sites did not need that much bandwidth. "We moved many to 64K bit/sec to 128K bit/sec ports, which saved quite a bit of money," she says.
Bernard also discovered a few sites that were in need of more bandwidth.
"We saved a significant sum last year and this year on downsizing," Bernard says. "We've saved five times the cost of the Orion product."
While Staples did not want to reveal specific dollar amounts, SolarWinds.Net Orion SL2000 software package starts at about $8,000. And while rightsizing bandwidth was an important goal for Staples, the retail store now also is better able to plan future network needs.
Before deploying the Orion product, Bernard's group would add capacity to the network when a new business application was going to be rolled out, without knowing if additional bandwidth was needed. The idea was to avoid potential performance problems.
"Now when someone says they want to deploy an application, we can pilot it before it goes out and also tell them that if they do roll this application out ...we will have to spend [this much] more for bandwidth," Bernard says.
Bernard says she constantly is looking at the Orion product, even when not conducting a network-wide utility analysis.
"In the right-hand corner of the [client], the tool reports your 25 most heavily used links," she says. "I keep an eye on that info, and if I continually see the same link pop up I download more detailed info on that connection. If capacity exceeds 70% utilization on a regular basis, it's time for an upgrade."
Looking for better performance
Although reducing costs is important to the South Carolina Army National Guard, improving network performance was the primary reason it deployed Visual's Visual UpTime management system.
"We were having difficulty at our small sites, especially those out in the boonies," says Captain David McNamee, network control center manager at the South Carolina Army National Guard.
McNamee supports a 60-node frame relay network. Although it's confined to one state, several carriers provide circuits because it spans many rural areas where only independent local exchange carriers offer service.
Network performance and reliability became an issue with many of the smaller providers. The regional National Guard not only supports data over its frame relay network, but also voice. Performance became such a problem that "it sounded as if some of our circuits were running across barbed wire," McNamee says. "We were also having trouble verifying our [service-level agreements]."
About two years ago, McNamee started working with Visual, but only in the past year has he gone though a full network deployment.
McNamee says he was looking for three things while selecting a network management tool: durable hardware, a versatile product that would let him troubleshoot remotely and to work with a respected vendor.
"The monitoring gear is often put in wiring closets that don't get a lot of attention" at the armories in each county, McNamee says. "With only two field technicians to serve the entire state, we needed a rugged device that didn't require these guys running around changing out power supplies and things like that."
His small staff also set the requirement for remote troubleshooting support. And because he was challenging his telecom providers, he wanted to team with a well-known vendor.
"Some of our providers were even offering to support the Visual devices, but I thought that was a little like asking the fox to watch the chicken coop," McNamee says.
McNamee's plan panned out.
"After we put the Visual probes in place, we would call our provider and say our Visual [probes] were indicating such and such problem. Then all we would hear was silence on the other end of the line, where we used to hear, 'It's not us,'" McNamee says.
Performance improved soon after because McNamee could prove exactly how a circuit was not performing up to par and also that the cause stemmed from the carrier.
McNamee also uses the information that he gathers from the probes to help justify current bandwidth and bandwidth increases that might be needed.
"Because of budget constraints, we're squeezing every penny out of every dollar," he says. "We have also been able to determine if Internet usage was affecting the [quality] of voice at a given location." This information has let the military organization advise users as to when to make downloads or file transfers to make better use of existing bandwidth.
Deploying frame relay management tools has benefited these three users, so you might wonder why everyone isn't using them. Brownlee Thomas, an analyst at Giga Information Group, says the there are two reasons: First, these management problems just now are becoming more sophisticated and therefore more useful; and second, these tools aren't free.