This is a deep dive into step one of our recent post on How to Improve Application Performance for Branches and Remote Users
In their November 2016 report, Digital Business and Cloud Demand New WAN Architectures, Gartner analysts Mark Fabbi and Neil Rickard state that, “to successfully serve the growing demand for public cloud-based digital business initiatives, infrastructure and operations leaders must fundamentally re-architect the WAN. Those who do not will inflict an expensive, subpar user experience on internal and external customers.”
Before you can even consider WAN architecture, the first step to improving the performance of your applications is knowing exactly where they are. If they’re in your data center, its easy, you already know where they are. But if they’re in the cloud, they could be almost anywhere.
While your apps may be in the cloud, the cloud itself is a bit of a misnomer, as every application ultimately resides on a physical server in a physical datacenter somewhere. This can get even more complicated depending whether the application is running as SaaS, on or incorporating PaaS, or running on IaaS. Of course, some SaaS and PaaS platforms run in their own datacenters, while some in third party collocations centers, and others on third party IaaS platforms like AWS and Google Cloud.
Adding one more layer of complexity is the fact that some applications are split, with front end, more static components, served over content delivery networks (which are more highly distributed) and back-end real-time application and data components served from the application provider’s datacenter or their IaaS platform provider.
So let’s take a look at a simple example. If you’re running a SaaS application, say Salesforce.com for example, Salesforce is running your application from one or more of its many “instances” globally. These instances are virtual clouds within its physical datacenters, some of which are operated on Salesforce infrastructure in collocation datacenters, and some within AWS.
Now lets say that in the U.S., you have users in St. Louis, your corporate datacenter, where your firewalls reside is in Atlanta, and your Salesforce instance is in San Jose. In this case, Unless you’ve deployed firewalls locally with every user group, will first go east to Atlanta over your network to clear security at your public IP gateway, then transit the public internet, cross country to San Jose, all along adding significant latency and degrading application performance. This is the “tromboning” effect I referred to in last week’s summary.
Thus the first step is to take an inventory of your applications, and for those apps not running on your own infrastructure, work with your application vendors, or an intermediary like Apcela, to determine where and how those applications are served. If you’re like most enterprises, with a deck of several hundred to several thousand applications, this is no small undertaking. In our experience, the good news is that, when it comes to application performance prioritization, the Pareto Principle typically applies.
Stay tuned next week for Step 2: Consider where future application deployment is likely to occur.
In the mean time, if you’d like to download a free copy of the Gartner report referenced above, you can get one from me, or online from our SD-WAN partner Viptela.