In Brief: Investigating MPLS Overutilization
IN PRACTICE | This article is part of a series of posts sharing examples of how AppNeta users have leveraged the service to solve performance problems.
Since business class networks are expensive, it’s important for the organizations that use them to have visibility into how they are used so they can make informed decisions. If the network becomes oversaturated, do they really need to purchase a larger connection? Or are there strategies they can implement to reduce consumption?
A global automotive component manufacturer uses AppNeta Performance Manager for exactly that insight. Each of their 50+ global locations uses an expensive MPLS connection for business-critical production traffic (load balanced between two routers at their larger sites) and Dynamic Multipoint VPN (DMVPN) via a local ISP for internet traffic and redundancy. Their corporate offices have multiple WiFi LANs, each dedicated to a different purpose (e.g., company computers, personal devices, and guest access).
Since bandwidth over the MPLS network is expensive and limited, local IT teams use AppNeta to react to overutilization by identifying the traffic in question and asking users to free up bandwidth by reducing their speed or rescheduling their task for after hours (whether it’s running a large backup or uploading/downloading large engineering files). While providing the local teams the visibility to react to individual instances of overutilization was an improvement, a Senior Network Engineer decided to take a more global approach to understand their network consumption. Let’s take a look at his approach.
1. What traffic is on my business class network? Is that traffic required?
Using AppNeta Performance Manager, he identified that a high volume of streaming media was utilizing the MPLS connection during business hours. Users were watching Netflix, YouTube and Facebook video.
2. Identify strategies to stop, minimize or divert non-critical traffic
While the company uses a third party vendor to block internet traffic they deem inappropriate, they did not want to explicitly block such streaming media services because some users require these applications for business purposes.
The Senior Network Engineer spent time in the Usage area of AppNeta Performance Manager, digging into which users were streaming at various locations during the day and discovered a commonality: the users were all joined to the Guest WiFi networks.
Armed with this information, the company decided to use the DMVPN tunnel to reroute Guest WiFi traffic to the internet using their cheaper local DSL connections. They immediately noticed an improvement: MPLS utilization dropped significantly at nearly every location, creating room for business-critical traffic and reducing the burden on local IT teams.
But what if he’d found a high volume of business-critical traffic?
In this example, the network engineer uncovered unexpected recreational traffic. But how could he have proceeded if the oversaturation was due to business important traffic? An important question to ask is whether the traffic volume fits the expectations for the specific application(s).
If the traffic volume is unexpectedly high, review the source(s) of the traffic to check for an increase in the use of that application. Alongside the details about who is generating traffic, AppNeta provides key application performance indicators including the actual retransmit rate, network latency, and application latency. If these values are abnormally high, it warrants an investigation into the application or network performance.
On the other hand, if the high volume of business-critical traffic is at an expected level, strategies should now be considered to allow additional bandwidth.
The value of understanding network traffic
A key value of AppNeta’s Usage monitoring is in being able to make informed, potentially cost-saving decisions about how to size and manage a network. Rather than buying more bandwidth if the network is at or near capacity, an organization can leverage AppNeta Performance Manager to understand what traffic is on their network and who is generating that traffic.