Cloud providers, and the IT leaders who choose them, are known for being somewhat obsessed with performance. That’s not a bad thing. Any organization that is about to entrust part of its IT infrastructure to a third party is going to need assurances: that costs will be reasonable, that security will be effective, and of course, that performance will be up to standards.
For cloud providers, speed, processors, and uptime are handy ways to quantify the value they offer and to show that they are keeping up their end of the deal. So cloud providers happily build dashboards for their clients to illustrate just how well things are running—the exact processor count, how much memory is being used, and so forth.
Here’s the problem: Quite often, cloud performance isn’t really determined by what happens in the cloud. Sometimes, users can’t even get to the cloud. Access to the cloud can be slow or diverted, by malice or by fat-fingering. In these cases, as in so many others, cloud performance, as perceived by the user, is actually determined by the so-called edge network that allows access to the cloud. “As much as half the performance of any database running in the cloud has to do with the edge,” says Michael Kane, a senior product marketing manager in Oracle’s Dyn global business unit.
Monitoring the Edge
If access to the cloud were closely monitored, issues of performance at the edge might not be so vexing. But even at many large enterprises, no one is watching closely.
It’s easy to see how this can cause trouble for users. An East Coast organization may have employees who travel regularly and use a cloud-based suite of office productivity tools and email. If an employee is in San Antonio, Texas, and tries to connect with that cloud, he or she is probably going be sent automatically to Houston. But if the Houston node is down, that employee is going to have difficulty getting access.
Even to access the office suite of products, to begin with, the company would need a reliable cloud-based domain name service (DNS) solution “That user is coming all the way across the internet, through multiple networks, and you need to see all of it,” says Kane. “If you can’t see it, you can’t tune it or avoid problem impacts.” When the employee eventually does get access, the performance will be subpar, and that will affect the entire transaction going through the cloud.
And the user is going to blame the hotel or café he or she is working out of, or whoever is providing the internet service, or worse, the targeted website owner.
Sometimes the trouble lies in passing content back to the user. Any particular site may be served by a content delivery network (CDN) responsible for streaming a video, bringing up a Microsoft Word document, or displaying a photo online. Again, says Kane, IT managers want to be able to figure out the problem in real time, and shift traffic if necessary.
Enterprises that are totally unaware of how traveling employees are using cloud services may still be doing a decent job of monitoring pathways between their own physical offices, but they tend to ignore or forget about staffers who show up where they’re not expected to be. An organization may monitor the path between a data center in London and an office in New York, but often doesn’t consider the users in Washington or Boston who also need access to New York. “You need to be able to have some method to watch that edge, and to watch that connection between your users and wherever your data might be,” says Kane. “Is it a valid path that gives them the connection they need, is it safe, and does it have good performance?”
The Internet as It Was Meant to Be
Virtual private networks (VPNs) are one solution, but they’re expensive and can be a single point of failure or bottleneck. IT managers want to be able to use the internet as it was meant to be used. The internet was designed to be redundant and self-healing; with active monitoring and managing of the edge, it can be. And that monitoring and management is available from cloud services providers such as Oracle. IT managers don’t have to install or implement anything differently.
Companies such as Facebook intuitively understand the importance of edge management. Their internet performance is everything to them, because they haven’t built their own infrastructure. For them, there is no fallback plan.
Even in the event of a cable cut that affects huge swaths of users—Kane points out that in San Francisco alone, there have been more than a dozen since 2014—the internet always provides another way to get users where they need to go. It may take a bit longer, but it can be done. The first step lies in creating high-performance, predictable, safe connections by which users can reach you. And that starts with the edge.
Kyle York is the general manager and vice president of business and product strategy for the Oracle Dyn Global Business Unit.