An IT Leadership Meetup Near the Golden Gate Bridge

Posted by on February 18th, 2015
March 18th, 2015

What happens when twenty IT leaders get together and discuss the changing role of networks in their organizations? Here’s a glimpse inside the room when Mohit Lad, CEO of ThousandEyes, was invited to lead a discussion on managing services in a world without network boundaries.

Three Things We Learned

We were invited to speak to a closed session of IT leaders from the Bay Area, representing companies such as Disney Interactive, eBay, HP, Safeway and Twitter. It was a closed door session (so I won’t be sharing anything sensitive here), but I will get into key themes that emerged. Let’s start at the end, with three things we learned from the session:

We were pleasantly surprised to get detailed questions on installation and setup. Most of the room had horror stories for deploying enterprise software – either taking too long or just going wrong. From the very beginning the group wanted to really understand the setup story. In our case, ThousandEyes is SaaS-based, so people can get up and running and looking at application delivery in a few minutes.

There were a number of questions on API integration – highlighting that products can’t be stand-alone these days. IT leaders want to know right away whether they can consolidate data in one place. This is different from 10 years ago and requires that every product should think API first. ThousandEyes RESTful APIs are documented, enabling applications to access performance metrics and insights captured by ThousandEyes.

CIOs want to get ahead of the problem and need to be proactive. There were a number of questions on alerting – catching notable events before they become major issues. Going further, looking to be put into a position where they have less problems to begin with. ThousandEyes alerting layers helps intercept issues before they impact applications and customers. For example a network alert can catch issues before they hit application availability. And better still, help prepare in advance before deploying new network strategies.

Three great takeaways, but what was the discussion that led to these? Here’s a quick synopsis of how it went.

How the World of Networks has Changed

The main question up for discussion was: How to manage services in a world where network boundaries are getting fuzzy? Organizations have moved from what used to be a controlled environment, where they owned the entire network, to a more complex environment, with network topologies spanning the organization, the Internet and the cloud service provider. “There’s now an aspect of the network we can’t see and can’t control,” observed one IT veteran. The room agreed: the Internet has become a critical part of corporate infrastructure – whether providing customers access to digital services or enabling employees to use cloud applications.

A New Set of Headaches. When you can’t see the entire environment, it’s very hard to troubleshoot issues when they occur. You’re no longer troubleshooting within your organization, but are working across organizations. This compounds the issue that silos represent in most IT groups.

Treat the Cause, Not the Symptom. Most solutions tell you the obvious. You have a fever, so you check the thermometer and it confirms that you have a fever. But it doesn’t tell you why. What if we can build a view that does not differentiate these different network segments, but treats them all as one coherent network? What if we can make the Internet look like it’s your own managed network? You’d be able to troubleshoot problems exponentially faster, monitor for issues before their impact is felt, and make better planning decisions when changing your enterprise architecture or when building new cloud applications. This is the premise of ThousandEyes – to radically simplify troubleshooting and performance management for modern networks.

Figure-1
Figure 1: How ThousandEyes collects data across networks.

Dissecting a Social Network Outage. Time to dig into a recent network outage using historical performance data captured from our public cloud agents deployed across the Internet. We used ThousandEyes to look at a world view of connectivity to Facebook and observed a couple of locations in Asia having issues reaching Facebook in the connection phase. We then jumped to the ThousandEyes Path Visualization, which presents a topological view of the Internet. We used it to immediately identify the hop and node in the regional site covering traffic to Asia affecting the application experience. Now that we’d identified the cause, we showed how with ThousandEyes, it’s easy to share a snapshot of what you’re seeing or an interactive link to Facebook to expedite resolution of the issue.

Benchmarking Public and Private Networks. One member of the group leading IT for a large multinational had also been experiencing performance issues from customers in Asia. We explained how he could deploy ThousandEyes Enterprise Agents within his own environment for the problem locations and compare this with ThousandEyes Cloud Agents – providing a public Internet view of the same region – enabling his team to benchmark and compare performance both inside and outside the enterprise network for the same location.

A number of the group voiced that ThousandEyes would be a great mechanism for “keeping your SaaS providers honest.” Mohit explained that it’s actually more expeditious using ThousandEyes collaboration framework to work with your SaaS providers (or vice versa) when identifying issues by sharing views into the network.

Digging Deeper – Routing and Hijacks. Another live dissection was for a large global manufacturer’s website. We looked a level deeper than previous examples at the control plane, searching for anything funky going on in the routing layer. We identified a BGP route that was flapping between two service providers, but for one of them, packets were being lost. From there, it was a few simple steps to identify the exact portion of the Internet causing the outage.

The issue of network security came up. ThousandEyes is used for DDoS mitigation, monitoring for DNS and BGP hijacks. We looked into a payments provider that was noticing latencies from Turkey to their payment portal. We immediately saw that the Turkey location was going to Netherlands, which was one of the termination points. From there, by looking deeper at BGP, we could see that the same block was being advertised from an ISP in Indonesia. This could be accidental or a malicious BGP hijack attempt – either way, DNS was resolving to the correct IP, but where was traffic terminating?

Figure-2
Figure 2: Monitoring for anomalous behavior.
Figure-3
Figure 3: BGP View During the Anomaly: Route Hijack.

It’s Never Too Late!

One of the group said that he could think of numerous times in his career where he’d wished he’d had ThousandEyes. Later in the discussion, Mohit responded with, “you wish you had this in the past. But it’s never too late!”

Planning is an important aspect for any IT group and one that we said we were seeing more of. For example, it’s helpful to know which instance of a SaaS provider will give you the best performance and whether an organization needs any architectural change to better connect to their SaaS provider. ThousandEyes enables this type of analysis and comes ready with eighteen data feeds for the most popular SaaS applications such as, such as NetSuite, Microsoft Office 365, Salesforce, Workday, Facebook, Twitter and more.

As more organizations rely on the Internet and cloud, networks are becoming more complex and we’re here to help. If you or your teams are spending far too much time responding to network issues, you can start monitoring your networks right away by signing up to ThousandEyes for free.

Processing...