Let’s play “spot the end user”

As I write this I am installed in the citizenM hotel at Amsterdam Schiphol airport. My 14 year old daughter is trapped in a deep gravitational well in her bed, so I’m going to have a bit of fun by sharing some snaps I took at the SDN NFV World Congress of the vendor stands.

The citizenM hotel chain — I’ve stayed in the splendid NYC one too — is to me a pinnacle of good user experience. It feels like every possible detail of your customer journey has been lovingly cared for and thought about. My sense is that the owners created a wallet-friendly designer hotel chain that they themselves would want to stay in.

So, how well is the telecoms industry focused on the experience it delivers? Would we want to stay in our own “telco hotel”? I accept that this conference is an event about the network internals: our “picks and shovels”. It is not an event about customer experience, network measurement, or service management conference.

That said, is this clever new technology helping to deliver a customer-focused outcome? My idea of “customer-focused” involves:

Is this asking too much? Well let’s take a slightly naughty and tongue-in-cheek look…

Before we begin, and important caveat: I’ve generally not spoken to these vendors, typically know next to nothing about their products, and am randomly taking snaps of the booths that happen to not have their marketing message blocked by someone standing in front of it. This is about taking the temperature of the industry, not the players or the event organisers.

128Technology are my “hot pick” for doing well in future. However, “your network is broken”, whilst painfully true, isn’t a happy message. We need to “fix the experience” and “fix the complexity”. Next stop: fix the marketing, as the product’s goodness is being hidden.

I am less convinced by this from IBM. Machine learning has its place in networking, but it carries a great danger: substituting an analysis of (likely transient) correlations for hard science and true causality. I would be asking questions about (hidden) risk if I were a prospective customer.

I like this more. Getting DevOps-y means adjusting the management methodology, and this is hinting that IBM are going to bring the best of IT management methods to networking.

It’s bit blurry on the left, but the edge and fog from Advantech is a proxy for delivering lower-latency compute to users. It’s heading down the yellow brick road to the magic kingdom of engineered experiences, and is only as far as the first scary forest.

I get the reduced cycle time bit, but what do you do and what makes you special? And how does it help me as a service provider differentiate? Time to say more clearly what the magic sauce tastes like…

Aha! Performance. One of my favourite words. I would be getting Windy at their booth if I had enough time. Although I would guess they are talking about performance of individual network elements, not complete systems or user experiences. But then again, I’m a bit of a demanding customer.

An elastic network from ECI hints at better matching of supply to demand. I would be spending time with them to qualify how well they characterise demand, how much control they have over supply, and how they assure the resulting user experience outcome.

This one from Intel. The future also involves Tuesdays, and we know what they are and they don’t make you dizzy. Unlike 5G.

Who knows who this “Huawei” start-up is, but they could go far. They seem to understand the future is an integration of the cloud compute with the network information photocopier. Bodes well. Maybe they might even crack cloud application access? Who knows…

Oooh, Ciena! I take one look at this and I am already having dreams of dinner with the sales and marketing VP and a long life of conjugal bliss. (Business only, I hasten to add, unless he’s really cute and single.) Experience, outcomesand analytics: that’s like wine, steak and chocolate mousse for dessert. Let’s make a date!

Um, err, because the problem our customers have is “vendors with not enough awards”? Let’s move on out of politeness…

VMware make digital axle grease. You can only get so user-centric with slippery gunge, even if it is essential to everything not seizing up.

The good news: Spirent are anchored in the place where demand is, the cloud, and are working out from there. That’s a user-centric start. The bad news: I dislike “smarter” as a term, as it’s devoid of human qualities and doesn’t mean anything technically. I suspect there’s some hidden goodness here, you just have to work to find it.

Let’s assume this quote is true. OK, you’ve taken your concept to market cycle time down. But will the concepts you can market be any better for users?

A future with Cisco involves no employees and all the screens switched off? Maybe they are the Digital Equipment Corp of networking after all… More seriously, the “any, any, any” is the core proposition of Internet Protocol (abstraction of connectivity), but it’s terrible at the isolation of performance and security. So I’d worry they might be solving the wrong problem really well.

I am sure Dell EMC have something unique to sell that makes a difference to users, given their joint history. Pray do tell us what it is!

This caught my eye. I would love to know what Inmanta’s performance model is if it is “end-to-end”. Would it survive scientific scrutiny? I did like the reference to “robustness” (although for early promotion to the next level, albeit at the price of general incomprehension, “antifragility” is what is actually needed). Still, robustness is only a hop, skip and a prance away from delivering resilience, which is something end users actually value.

Two words here that make me swoon: “hypervisor” and “interoperable”. I could geek out with this lot, although I am not sure what it does for users. It’s cool anyhow.

Aha! A real, live service provider. What a change in language and outlook! “Reliable”, “consistent”, “guaranteed”, all wrapped up in lovely locality and ready to meet the “most demanding”. These people at Colt are at least talking the CIO’s language. So why aren’t their suppliers, too?

I so want to love this lot at the world’s hardest-to-pronounce vendor for having testing and measurement, but it seems strangely lacking in words anchored in the end user reality. No “service”, “quality”, “experience”, “assured”, “risk”, “accurate” (or similar)… you get the idea. But it’s DevOps, so something is going the right way.

Why do I have to throw away 20% of the resource? Can’t I use it all? And why would I care about a “SR-IOV optimisation”?

Wots a “60G Ericsson vEPG”? Is it when a dozen 5G vEPGs get together for a party? I’ve been in telecoms for 15 years, worked for two telcos, been a senior analyst, and I have no idea what this means. I couldn’t begin to tell you what it might do for a user, even if I used my most polished Oxbridge make-it-up-on-the-spot skills. Maybe it’s just me? Or is it them?

Better news from Ericsson, who understand the demand is getting cloudy and we are in a distributed computing paradigm. Maybe there’s hope after all…

---

So, a few words of summary. This is an industry that’s still finding its way. SDN is doing “low frequency trading” of resources on a bandwidth model using averaged scalar metrics. The end user experiences the continuous passing of the instantaneous performance of the network, which the SDN orchestration doesn’t model or control.

Until the industry moves to high-fidelity metrics and high-frequency trading of resources, it will only have a very weak influence over the end user experience. Unsurprisingly, the current marketing messages reflect the technical disconnect from the end user and their experience.

There are signs that the cloud, dynamic resource trading, and experience-centric outcomes are part of the SDN NFV industry landscape. It is definitely working hard on “faster cycle time” for products. Some of them may even be cheaper to deliver with automation. Just the “better for users” seems to be low in relative emphasis.

Now it is time to watch who prospers. Will those with a more user-centric view lead future growth and take up positions of market leadership? Or are the needs really and truly all deep in the internal guts of the network to squeeze out network architect and engineer labour costs?