- 1 Red Hat Enterprise Linux 7.2 Enters Beta with Improved Container Support
- 2 VMware CEO Pat Gelsinger Gives VMworld 5 Imperatives for Success
- 3 VMware vSphere Integrated Containers Previewed at VMworld
- 4 Worldwide Server Revenues Top $13.5 Billion in 2Q15
- 5 Blue Box OpenStack Lands on IBM Softlayer Servers
Bandwidth Not Always the Reason for Slowdowns
Historically, the obvious solution to any slowdown has been to add more bandwidth. Yet sometimes all it really does is delay the problem at a higher cost. The culprit of the situation may actually be bandwidth abuse, mis-configuration, application overload, disk constraints, overprovisioning of virtual servers or any number of other reasons.Find the root cause of that slow network before you throw more money at increasing bandwidth.
"Daily network monitoring can play an important role in determining the source," said Stephen Brown, product marketing manager at Network Instruments LLC of Minnetonka, Minn.
Take the case of a company suffering from unusually high network utilization at a remote location. It turned out that only three IP addresses were consuming as much bandwidth as the rest of the users in that location combined. Investigation on network monitoring data found them running rogue online backups of their laptops during business hours. In this case, the solution was to police users, not add bandwidth.
Voice over IP (VoIP) implementations are a common area where bandwidth is seen as the solution to all challenges. To prevent a reduction in voice quality or address user complaints, more bandwidth is the obvious step. But is it really addressing the cause?
"We usually see issues of people throwing bandwidth at VoIP and unified communications problems," said Brown. "VoIP isn't the type of technology that network managers should attempt to implement without doing a little legwork up front."
A common mistake with VoIP, for example, is to implement it rapidly and then attempt to troubleshoot issues on the fly. Network managers typically anticipate that adding VoIP will increase bandwidth needs and, as a result, automatically upgrade their bandwidth capacity. Brown believes, however, that the increase in bandwidth is often a waste of money, as it doesn't address the cause of the performance problems.
Brown offers several suggestions as to how companies can gain more insight into their networks and better control VoIP, unified communications and other networking projects.
Utilize Quality of Service (QoS)
QoS is imperative with business-critical applications like unified communications. Like VoIP, unified communications applications need consistent network performance, as problems from other applications easily impact them.
"The way to ensure access to bandwidth for critical applications is by using QoS tools," said Brown.
Another key step an engineer can take is to conduct a pre-deployment assessment of the network. This makes it possible to understand current bandwidth load and existing network and application demands before deploying anything.
"In the case of VoIP, instead of guessing how it will impact the network, the best approach is to use a network analyzer to conduct a pre-deployment assessment of the network," he said. "The assessment allows you to prepare your network for the added VoIP traffic, identify and resolve obstacles that would get in the way during implementation, and create performance baselines that can be used to evaluate post-deployment success."
Pre-deployment assessments present an opportunity to discuss VoIP expectations with managers and plan appropriately for long-term network growth. All of this, of course, is much easier to do prior to deployment rather than after the fact.
Saving Money and Bandwidth Through Daily Monitoring
One Network Instruments customer, Sussex Health Informatics (HIS), for example, had a network of 200 remote sites and 25,000 users. It needed to upgrade its network to support an image archiving system and replace its existing WAN with a newer, more resilient one. In making this multi-million dollar investment, HIS also needed to be able to troubleshoot future roll outs and guarantee delivery of network services.
The image archival system would replace the healthcare provider's traditional X-ray machines. The system vendor told the team it would need to invest in Gigabit WAN links to ensure performance. However, using a network analyzer (Network Instruments' Observer), HIS was able to test the network and determine that the existing 100 MB links were largely sufficient. Gigabit WAN links were necessary on only rare occasions. This saved the organization a substantial amount during the upgrade.
"Long-term application performance monitoring and trending is critical for establishing baselines of normal network performance," said Brown. "You should rely on your network analyzer daily, rather than pulling it off the shelf only during an emergency."
This article was originally published on Enterprise IT Planet.