If LinkedIn profiles are any indication, User Experience (frequently shortened to UX) is the new orange. Indeed, across all manners of technology, there is an increasing focus on improving user experience. Driven in part by Apple’s success on the consumer side, it would appear that IT infrastructure vendors are getting in on the action. In the quest to simplify our collective lives and differentiate in a space more defined by cost than capability, the user is taking a more prominent role.
As it should be.
Networking and UX
This is especially true in networking. The truth about all but the highest-end markets is that from a performance perspective, there are multiple options that are good enough. The differentiating and most compelling points for most people are less about capability and more about cost. For now, the focus remains on reducing CapEx, but once all the vendors settle out around the same price (everyone is using more or less the same hardware), eyes will turn more towards OpEx. And that will bring into play a lot more of the operational experience in deploying and managing a network.
First, while decision criteria will initially settle on cost, make no mistake about it: the actual differentiation is bigger than just cost. The overall user experience—from architecting to procuring to deploying to managing—will become a means of distinguishing solutions. When the window for cost narrows, it actually takes cost off the table and makes other elements of the product even more important.
As a strategic aside, this is why some of the more raw DIY solutions are going to struggle a bit in many parts of the market. The downward pressure they exert on price is effective, but once the more entrenched players come down, the difference in price is no longer compelling. This leaves the primary differentiation as capabilities, global support, and user experience, all of which tilt towards the legacy companies – at least in the short to medium term.
State of the industry
If user experience is poised to be a big difference maker, what does the state of the industry look like currently?
In a word, sorry. IT generally, and networking especially, is in a sorry state when it comes to overall user experience. Most networking solutions are provisioned and managed via device-by-device commands. In the best of cases, this is a repetitive task that can be automated; in the worst cases, the commands are esoteric, highly contextual, and vary in behavior depending on conditions.
We exist in an inside-out industry. User experience is the last thing that vendors think about when determining direction. Roadmap planning in any of the major vendors is an exercise in spreadsheet management. Individual feature requests are fielded by sales teams and filtered to product management teams. There is typically a rough sort on business case (how many deals are riding on a particular feature?). From there, the roadmap is reviewed, tweaked, and published.
Of course there are user-experience champions at every company. They fight the good fight. But there are only so many places they can insert, and the roadmap is dominated by people who are not so enlightened.
In a user-experience-centric world, the starting point wouldn’t be the feature. At least not in its current state. Each feature would be in support of some objective. That objective would map to a set of workflows required to achieve it. And those workflows would all have properties.
- Every workflow has a trigger. What initiates the workflow? Is it someone keying in a command? Someone clicking on a GUI? Maybe an API call?
- Based on the trigger, what are the likely contexts for the workflow? If the trigger is based on someone logged into a discrete device, then the CLI might be an appropriate context. If the trigger is an API call, the context might be entirely different.
- How do you know when a workflow is complete? Is there some state that is captured?
- Most workflows will exist alongside others in workflow chains. What are the preceding and following workflows? This helps identify possible improvements in automation on the system side. For example, automatically pinging across a link after setting up a BGP connection would be a fairly trivial way of adding some validation to a provisioning workflow.
- Some workflows have dependencies on other workflows. Mapping those dependencies out is an interesting way to help simplify user interaction. If, for instance, a workflow is predicated on the existence of a policy definition, automatically teeing up policy creation is an interesting workflow element.
There are probably a dozen other properties of workflows. The point here is not that these are the right properties but rather that adding a new feature ought to begin with the workflow in mind.
UX and architecture
In networking, we equate UX to UI. And then we throw a CLI on virtually everything. A user’s experience goes well beyond just the interface into which a user keys a command. If we really thought about workflows first, we would do more than add configuration knobs. We would consider validation and troubleshooting as natural extensions of workflow, and make these easier to do.
The unfortunate byproduct of our collective approach to user experience is that we do product architecture first. The challenge here is that for user experience to be truly effective, it ought to inform the underlying architecture. As it stands, user experience is a consumer of that architecture, with very little (read: no) input in the vast majority of cases.
Realize that the networking industry has been architecting products and solutions from the inside out for decades. Developers are in a tough spot. They have to make magic happen by typing into a terminal window. The focus is function first, form later. And this leaves user experience relegated to a tack-on after the product has already been built. At that point, how much can you do anyway? This is why UX has equalled UI for so long. Without architectural input, the best we can hope for is a slightly better configuration knob.
The bottom line
If networking really is on the cusp of changing over, user experience has to be an input into the design process. This is a radical change in how products are conceived, built, and ultimately brought to market. Do not underestimate the difficulty in transitioning an organization from feature-centric to user-experience-centric. Most established companies will fail to make this transition (if they even try at all).
The biggest impact of startups in the networking space might be a refocusing of efforts around the thing that we seem to have lost amidst all the technological upheaval over the last few decades: the user. And wouldn’t that be a glorious thing?