Product News
Announcing Cloud Insights for Amazon Web Services

Industry

Modern Wide Area Network Monitoring: Migrating to DIA

By Tim Hale
| | 11 min read

Summary


In my previous post in this mini-series, I covered why so many customers we’re talking to are considering migrating away from MPLS and the journey to SD-WAN. In this follow-up post on wide area network monitoring, we’ll look at the growing pains associated with migrating to Direct Internet Access (DIA).

Losing Visibility

Regardless of whether you choose to implement a more basic branch DIA architecture using per-site policy routing, or an SD-WAN, moving away from MPLS networking creates a major challenge and one which all of the customers I’ve been speaking to recently have been wrestling with. Namely; if you don’t own the infrastructure and the application is provided from someone else’s data center, how do you get the right wide area network monitoring data to manage performance and deliver a good user experience?

Traditional MPLS network circuits are managed by one provider who is responsible for service levels and responds to all support tickets. The physical infrastructure of the rest of the network is wholly owned by the enterprise therefore engineers have access to the hardware and can more easily perform network diagnostics using passive monitoring techniques like packet capture, SNMP and NetFlow to analyse and investigate issues. They can also define their own internal SLAs for these parts of the network.

The increased uptake of cloud services and the resulting network transformation means a huge shift in how networks are managed and monitored. Elements of the legacy corporate network still remain, with some applications staying in the data center and some sites keeping their MPLS connectivity, but there are now large blind spots for IT teams, since they simply can’t access or monitor the infrastructure of public Internet and cloud providers. In this hybrid environment old monitoring technologies still have their place but companies have to ask themselves if they can effectively troubleshoot issues on the public Internet and with 3rd party networks.

Three Reasons Why DIA Transformations Experience Failures

From what I’ve seen there are three key reasons why customer branch router upgrades and DIA network migration projects fail.

1. Unclear Success Metrics. It’s hard to succeed if you don’t have agreement on what success looks like. A staggering number of IT projects are considered failures by business leaders even when the project team considers the project a success. This is why the metrics for success in a network transformation project need to be carefully considered. We’ve seen DIA projects where success metrics weren’t clearly defined or linked back to business objectives. It’s worth asking questions such as:

  • Other than cost savings, what are your metrics for success?
  • How will you benchmark those metrics?
  • How will you present data to non-technical colleagues?

2. Unrealistic Expectations and Unfunded Increases in Support Requirements. The Internet is not the relatively boring, predictable and measurable WAN service that you’ve loved or hated for the last fifteen years. That might seem like a really obvious statement, but those focused primarily on the cost-savings angle of DIA may be exercising some level of convenient thinking, where the complexities of the Internet aren’t a factor.

One of the hidden pain points of DIA that emerges as a result of increased reliance on the public Internet for application delivery is an increase in support tickets. These Internet-related issues can include regional ISP outages, incorrectly prioritised voice data or sub optimal routing. This can stretch an already busy IT team, especially if they’re used to simply handing tickets off to an MPLS provider or owning all the network hardware themselves. Either headcount needs to go up or triage time needs to be reduced.

So it’s worth counting the cost of the switch in terms of the time it will take to troubleshoot issues in the “new normal” environment of the Internet and consider questions like:

  • Do you require extra headcount?
  • What is the current Mean Time To Repair/Restore and can it be improved?
  • Do you have the visibility needed to keep the handling of these support tickets manageable, especially given that service desks are often understaffed relative to demand.

3. Ongoing or prolonged issues and outages. Project owners need to consider and manage the internal perception of the project. Network transformations are always high profile and not responding quickly and effectively to issues on a new network can lead to a negative perception of the project.

The network should be benchmarked before, during and after changes to ensure real data can be used to demonstrate performance improvements and prove innocence when events that are out of the control of the network team lead to poor user experience.

Modern Wide Area Network Monitoring Helps Your DIA Migration

The modern wide area network has to change. Cloud and mobile adoption and the integrated ecosystem around them mean the traditional MPLS based network no longer makes much sense. The center of gravity for the WAN is shifting to the Internet, which means you need new wide area network monitoring capabilities. That’s not to say that traditional wide area network monitoring technologies don’t still have a role to play. However, when architecting for the modern WAN, companies must ensure that even if they don’t own the networks, they can still own the outcome. And you should consider questions such as:

  • Will you be alerted to issues before users create support tickets?
  • Are you able to triage and provide solutions to issues on the public Internet and in third party networks?
  • Do you have visibility across your entire WAN?

At ThousandEyes, we exist to help IT teams see, understand and improve connected experience for all their users. We do this via a modern take on wide area network monitoring delivered via a cloud service which can blend data gathered both actively and passively from desktop, site-based and cloud-based software agents, and that gives you full visibility of application health linked to network performance, regardless of whether you own the network or not.

If you’d like to get a bigger picture take on how the enterprise WAN is becoming more Internet-centric and how Network Intelligence helps that transformation, check out our Network Intelligence for the Modern Enterprise WAN eBook. If you know you need Network Intelligence to deal with your WAN transformation, request a demo or start a free trial.

Subscribe to the ThousandEyes Blog

Stay connected with blog updates and outage reports delivered while they're still fresh.

Upgrade your browser to view our website properly.

Please download the latest version of Chrome, Firefox or Microsoft Edge.

More detail