Application Performance: User Experience
CTO, Kerv Connect|Kerv Connect
Published 01/11/22 under:
As users and consumers of IT, we have a very external facing view of what needs to hang together for everything that we do on a day to day basis to work.
What I mean by that is that as a consumer of an application, sat in front of our laptop, or on our mobile phone we have a very external view of the experience that we’re getting. We understand whether it’s a pleasant experience, we understand whether or not it is performing in terms of responding, when we are pressing buttons on websites, and as we engage with different kinds of content.
Increasingly, the sort of applications that we’re engaging with are what’s called distributed applications which means that they are not controlled in one place; they’ve got arms and legs everywhere. An example is an e-commerce site which is essentially an application, you’re carrying out certain things, you’re buying certain things, and you’re pressing certain buttons. Or it could be a business application that automates a process that you do at work, it could be something like a video call, or it could be a CRM system.
Those are all application experiences, that from time to time, depending on your role, or your job, or what your personal environment looks like, are things that we engage with. We have a perception as to whether or not that’s a good experience or a bad experience. From a networking standpoint, that typically manifests as, “Ugh, the network is slow today.” if it’s a poor experience, or “the Wi Fi is rubbish here.” situation. In short, the networking industry is a really good example of an industry where you are guilty until proven innocent.
The challenge is that in reality there’s a lot more going on there than just the network. The network is about providing connectivity, but in order for an application to work over that medium there is a lot to orchestrate. For instance, you have to be authenticated to use the application that you’ve got, the systems that do this could be slow and not performing well. Those aren’t necessarily a network function, but are a critical path of what your experience looks and feels like. For the user this manifests as ‘the network is slow’ or ‘the WiFi is rubbish’, because you don’t know that it’s the authentication medium that is struggling to respond.
The thing that renders a particular part of an application in a web format, within a browser, could be running into issues. That’s much more of a generic IT problem than it is a network problem, and those things are siloed across large organisations; you don’t necessarily know that there’s a problem in terms of the particular area. Problems can manifest in multiple different technology silos impacting application performance and end user experience. The guilt tends to lie at the network; hence the reason why you’re guilty until proven innocent. It’s always the network’s fault until you prove it’s not. So a big part of our job is around ‘it’s not us it’s you’ situations.
SD-WAN helps in that model because when you are running into issues, it will automate the navigation around those issues. So, if something is not responding in a timely fashion, rather than trying to find out what the issue is, which is all you really had in terms of options to address the problem before, it will just try to take another route. If you’re dropping traffic as a consequence of trying to get to a particular location, the cause of that might not be a network issue but you’re dropping traffic nonetheless. If the problem is intermittent it will take remedial action by automatically sending more than one copy of the data to increase the chances of it making a successful journey.
First of all SD-WAN helps in navigating and remediating against problems and the problem is a really, really complicated one. What we’re also seeing is the integration of SD-WAN into a series of technologies that get bundled into this thing called AI ops; artificial intelligence operations. What that’s trying to do is to take a much more holistic view of the end user experience. What we’re used to in the networking world is that there are these component parts of a network, and we need to monitor that all of those component parts are working well and within parameters. But that’s like having lights up in the ceiling that casts light down so you can see on top of all of the component parts.
What’s becoming much more prevalent these days is the ability to be able to shine a light through the conversation you’re having on a per application basis. So it’s much more granular in terms of what that experience is. What that gives is full context of what the end user is experiencing at every part of that journey, and because it can use artificial intelligence techniques to take data from all of the component parts, it can start to correlate where your issues are at. As that develops, and this is on the future roadmap for things like SD-WAN, you become more aware of where that’s going to happen, SD-WAN can take proactive actions to be able to navigate around those issues. SD-WAN is really effective at improving end user application experience. It’s not necessarily solving the problems that are happening, but it’s giving you visibility that they’re happening, and automatically remediating around them if it has options to be able to remediate around them.
It’s actually protecting the end user experience, even though issues are occurring, and the issue that you used to have in a network world was that something would have to go completely down in order to move on to an alternative route. For example: if I had two links into a building, one would need to break completely in order to flip over onto the other link. Well, the issue quite often is not that the link is down, it’s that the link is degraded, or the application experience over that link is degraded. SD-WAN understands the context of that, whereas traditional WANs don’t. So the tagline that we use for that is ‘degraded is the new down’.
Degraded can in fact be even worse than it being down. If it doesn’t work at all, it doesn’t work, but if you’re frustrated in terms of how long something is taking, that can be infuriating. You’re pressing a button and not getting a response, and that translates into wasted time and lost man hours. You can imagine where this starts to manifest in terms of real losses for businesses. So this is how it plays into application performance and user experience.
It’s an infinitely complex problem. If you actually laid out from end to end all of the component parts of an application, you’d be really surprised where that goes. We saw a visualisation quite recently of a Microsoft Dynamics CRM system, which is part of the Microsoft Software as a Service. It breaks out into services that are delivered for nearly 37 different physical locations in Europe so you can imagine all the network connections that are needed and all of the component parts that need to be orchestrated in order to deliver that visualisation of that application. Some of those are on the critical path that you’ve been able to do your work and some of those aren’t.
In order for that application to work properly, all 37 locations, all connections into those data centers need to be working in an orchestrated way. If one needs to load in order for you to complete the workflow that you need to do and it’s struggling, then you’ve got an impact in terms of application performance and end user experience. It’s hugely complicated because applications don’t tend to sit in one location anymore. Some of the data is sat within a corporate data center, some is in AWS in the public cloud, some is in Microsoft Azure. It’s everywhere, the world has become hugely fragmented. Quite often it’s not an infrastructure that you either own or control anymore; it can be a problem with Microsoft’s network, it can be a problem with Amazon’s network, it can be a problem in Google’s. This is the sort of mess that we try to unpick when you run into issues, which is why we employ the people that we do because when there are issues, these are the guys that have to track down where the problem is and it’s infinitely complicated.