This time in Thought-leaders on Emerging Technologies we spoke to Mohammed Khan who is a specialist in Telecommunications and has 15 years of experience working at the forefront of Network Technologies. He previously worked at Juniper, followed by Huawei where he was the CTO of SDN/NFV and Cloud Solutions in Shenzhen, China. When he moved back from China to the UK he continued his extensive career as CTO Virtualisation and Cloud EMEA at Netcracker.
Keep reading to find out more about SDN/NFV, the future of networking, Network Engineers and Mohammed’s one piece of advice for you
When did you first hear about SDN/NFV?
Back in the day in 1997 I worked at BT and I was working on the concept of SDN about distributing computing. We wanted to look at how we could make the underlying network a lot more intelligent and that is where we started looking at open API systems. We worked with the lights of Cisco, Microsoft and ATT. Everyone talked about these open systems. But then it kind of died because we started to get into routers, bigger routers and all the rest of it.
In about 2008 one of the objects I had when I was working at Tellabs was to see how we could build a router based on a Linux environment. Once we build our Linux environment, we had to find out how much we could keep on the router. Such as the line cards and how much could we move the intelligence somewhere else. I remember looking at it thinking this is looking neat.
The first time I really heard about SDN and NFV must have been in about 2010. I got on to these working groups in the ITF. The SDN concept makes sense, because the problem with routers is that when you let a router make a decision it can only see the routers that are directly connected to it. It can’t make a holistic view of what the network is saying.
Where routers on a hop by hop basis really struggle with. Regardless of what anybody says about the visibility and connectivity, if you leave it at router level it’s very difficult. Because it’s forever changing.
SDN takes the control plain higher. It allows all of those interactions to take place and the machine runs through its algorithms and what it then does is at this time stamp this is how it is looking but I believe this connection could die because it is already running at 60 or 70%. So, it can preamp and run another route. That’s what SDN is.
What where your first thoughts?
I liked it! I knew there were limitations but I liked the idea. The ability to take the control and decision making away from the router and put it on a server is what interests me. And make sure the server has more processing power, more RAM and more storage then it ever is going to need on a router. The key is the processing power, how quickly can it run through the algorithms of the if/then scenarios to tell you what’s happening down below. It not only tells you exactly what’s there but it also gives you an insight what can happen on previous analytics that it has. Pre-empting is how things should be.
Strengths and weaknesses of SDN?
The strengths are that you have a lot more flexibility and efficiency in your underlying environment.
“I don’t even call it a network anymore, because it’s an environment”
And it’s also far more agile when you have to deal with things.
The opportunities in this industry is a strength as well. anyone who can write code and is agile enough and has a good business strategy can get into this market now. It’s no longer a closed market.
The weaknesses are that you need to have really good control mechanisms. You need to have some sort of certification that this is how the environment underneath of the intelligence is at that time and space.
The other weakness is security. You open up your network and everyone can connect to it and everyone can work on it. It’s not just having a firewall stuck in there, it’s more than that. You need to have a complete guide by which your environment can connect to another environment. It’s complete revamping the way you do your business regarding your people, your processes, your automation, your changes and your responsibilities. Everything alters. At the beginning when you put this in your environment you still have to press the button “yes” that’s ok, do it. Overtime it becomes automated. Who makes that decision that it seems to be a reoccurrence, we can let it do it automatically? So, it’s not about a firewall. Most hacking gets done by people who used to work in the company. It’s an external problem. It’s a hole that has been left there by someone. What the others do is just explore that hole.
So, the security is one weaknesses and so is the change management. You will need to do a lot of re-training, you need new skills, a gap analysis on your skills etc. All of that needs to be looked at as well.
Another weakness is that if you are once the incumbent in building platforms, you are going to be under the threat of someone else. In the way it is now, you are going to look at someone who is doing SD-WAN, such as Silver Peak. All these people come from nowhere because there is an opportunity to make money.
What do you like most about the industry?
It’s a lot more buoyant than it was 5 to 10 years ago. It’s changing, the skillset and mindset are changing. And it’s a challenge. The challenge is more on the players then it’s on the vendor side. With the players I mean companies like BT, SKY, Amazon Web Services and Vodafone. Because all of them has a legacy system they have to do something with. They can’t just switch it off and switch it on again in two years. None of them come with a green field. They all come with something they have to migrate across. It’s a huge challenge for them.
Who do you see as the biggest innovators in the SDN space?
The small software houses. The once who have a really good business plan and business idea. Those are the once that can clean up. Because this operation of being a very linear service provider to being a cloud player will go into marketing, strategy, HR, operations, engineering and maintenance. So, anyone who comes with something innovative, something that will help make it more efficient, automated, agile and a better user experience will win. I think that’s great.
A lot of companies in SD-WAN are really good at it because they really show what the product can do.
Who do you see as the biggest adaptors?
The biggest adaptors are companies like Amazon Web services, Facebook, Google Cloud and Microsoft Azure. All the companies who have something to build beyond their original core product. Who would’ve thought 15 years ago Amazon would be into building Cloud Environments? They couldn’t even deliver a book on time, now they are building you a Cloud Environment.
Of the traditional companies I would say the ones who are really leading this are AT&T, NTT docomo, Deutsche Telecom and Vodafone. The other ones are all getting there, but I would say these are the for runners. The reason these are not moving as fast is because take for example Vodafone, they have around half a billion customers. They can’t move at the same speed as for example Amazon Web Services. At the end of the day Amazon Web Services build the environments in the cloud. They are not giving you the connectivity. Where These guys have to build all of it from the ground up. Deutsche Telecom are the only ones I know of that can build you a virtual machine in the cloud alongside Google.
“Have a look at Bitnami. It’s the best cloud environment to build environments. It’s a bit like GitHub. The only cloud server environment that you will get here are Amazon Web services, google Cloud and Microsoft Azure and the only service provider is Deutsche Telekom.
Deutsche Telekom are very good at it and are way ahead of the competition. German efficiency”
Where do you see the future of networking going?
It’s going to be in the Cloud. There are still some constraints about how much of the cloud can be open. There are certain things such as optical and call routers which are going to stay around for a long time. It’s very hard for them to go on SDN/NFV because of the jobs that they do. But the rest can be open. And the other thing is that the environment is not going to be linear anymore.
What implications does this have on our modern day “Network Engineers”
It’s all going to alter. If you don’t know how to code you are pretty much done.
“You will be in trouble if you can’t code”
Look at the job market today. If you apply for a job but don’t know about SDN/NFV, Python and scripting you’re in trouble. It’s all going to be automated
Will the need for them go away?
No, it won’t go away, it will just come with a new hat on. We will still need them, but then just with a new hat. Instead of the old days where you used to get into a router and find out what it’s doing, maybe now you will have to debug a piece of script.
Look at it like flying. There is auto pilot, but it doesn’t mean pilots don’t have a job anymore and that the need has gone away. Auto pilot made their job easier.
What is your biggest difficulty when recruiting?
You always ended up recruiting based on what you’ve deployed. Which is absolute nonsense. This is what I like about the open. It means its more agile and your skillset will be different. It’s entirely up to you if you want to learn it or not. You don’t want to recruit someone who just does one thing.
If you could give 1 piece of advice to a Network Engineer what would it be?
Be open minded to change, it’s going to come. You might as well change.
Also, start building websites. Because those skills are almost the same as the ones you need. All the scripts are going to run through an apache webserver.