Can VNFs match the performance of PNFs?
At some point before the procurement and sales teams give each other high fives for a technology purchase well negotiated, I always hope somebody had the courage to call in my favorite kind of techie: the ShowMe Skeptic. This guy or gal is never popular. Seldom gets invited to the party. That's because she warns you that you're about to make a mistake. But when you're about to ink a multi-million dollar deal, that last thing you want to hear is that you're about to make a mistake!
The temptation to ignore her warning is often too strong to resist. That's just human nature. We want what we want, and we don't want to wait. Particularly when dealing with technology.
So we sign the deal, we deploy the technology, and guess what? Our skeptic was right. Do we thank her? No, we hide from her. Because we know she was right. And we're too embarrassed to show our face.
So yeah, the ShowMe Skeptic, even though she would do nothing but help us if we invited her to the party, does not get invited to the party. Which is sad. Because she's a whole lot of fun.
End-to-end VNF performance
Take notion of VNF performance, for instance. When we talk about digital transformation in telco-grade network services, we focus on the business agility we will gain, on the cost savings, the operational efficiency, the potential for 5G/ We dream about the ways automation will simplify management and make us even more reliable. It's pretty heady stuff. It's exciting. It's promising.
Except to our skeptic, who wonders, when the software-defined infrastructure hits the pooled hardware resources, how well it will all perform (That's a dim play on "when the rubber meets the road," so I hope you saw what I was trying to do there.).
If someone did have the courage to invite our skeptic to the right meetings, she would have voiced that concern. The response of the sales team would be to wave its hands in the air and say that in our new software-defined paradigm, the notion of performance has to be re-interpreted. It's more than just Gbps through a wire. It's operational efficiency, it's business agility, it's The Edge, it's rapidly-rebooting micro services, it's rapid deployment of containerized environments. It's IoT. It's BigData. It's 5G.
To which our skeptic will cooly respond with "Do you have any metrics you can share with me?"
"Of course we do!" the sales team shouts, knocking over a couple of laptops in its rush to get to the facts. "Our vIMS service is 30% faster than its IMS equivalent hosted on dedicated hardware," they read from a presentation. "And our virtual version of VoLTE is faster, too."
"How about your content distribution network? Or your edge routers? How do those perform?"
The room goes quite and people look at each other uncomfortably. "Uh, um, well," the team lead says, "you need to realize that with our new paradigm, the notion of performance has to be re-interpreted."
And so the conversation goes. Which is why our skeptic is never popular.
However, when it comes to the performance of VNFs, the skeptic should always be invited to the meeting. Because she will ask critically important questions. Yes, as our customers have discovered, not only do telco-grade network functions hosted on a cloud infrastructure deliver the business agility and operational efficiency that we promised they would, but our virtual network functions are, indeed, faster than their physical equivalents.
But not all of them. Or perhaps not yet. Or perhaps it takes a little work. And a different way of approaching the problem. And maybe it makes sense to keep some physical network functions running for a while longer, while you transform the others into virtual. And the only way to know which is what and which way to go, is to invite the skeptic. Because she will ask the questions that reveal the real issues. And help you make a smart decision.
We wrote this eBrief with her in mind. It's a short, but honest discussion about how VNFs perform relative to PNFs. With recommendations for how to tune your VNFs so they perform as well as they can. Download it, show it to your skeptic, and ask her to send us any questions she still has.Download eBrief chapter 7
Also, listen to this podcast where Mats Johansson, Head of Marketing for NFV, shares his view on how to take NFV from trial to commercial operation, giving insights from the first live NFV networks. Enjoy!