Future Of Network Engineering

Bardlebee

Hey guys!

As you know I've been working hard on my CCIE, like many of us here. I would like to though, take a minute to ask you guys what your thoughts are on the future of network engineering. I am talking perhaps 5 - 20 years in the future.

All the hours I am putting into the CCIE has helped me immensely at my job and what I do and eventually I do hope it gets me my number. But I'm also worried (as I worry all the time) if all this work for a CCIE number will hold its muster in 10-20 years. Even 5 years?

I know many people are concerned about SDN (Software Defined Networks) as some boogie man that will take our jobs, but I wanted to know what you guys thought of that. I am basically looking to future proof myself in a way. After the CCIE should I focus my studies on learning Python and API management of network devices? Or do you think SDN is just a bullshit term that will never take off to the point where it puts half of us out of a job?

What should I focus on to remain a Senior Network Engineer and be at the top of my game for the impending future is what I am concerned about and also interested to hear how you guys feel. Personally I feel I need to learn SDN, for the just in case factor.

Thoughts?

hurricane1091

You should take what I say with a grain of salt, but I did a webex with Cisco for their SDN stuff and I cannot fathom that being something that takes away what we do.

Maybe one day there will be a better way for things to be done, but think about some things that we do. Turning up a new circuit is literally going to involve boots on the ground and then configuring the interface. IP schemes and VLANs are going to need to be drawn out and planned still. IGPs are not going to magically turn up by themselves, etc etc. I hope I am not naive, but from the webex I had I was not sold on anything. 100% new equipment was needed and it is expensive obviously. Training needs to occur then too. Maybe they are working on new things now, I do not know for sure. Their whole model was based on this APIC controller, which quite frankly seems like it came into play when developers created new programs to help auto configure some networking stuff.

I'm not saying the future will not be different, I am just saying in the short term 5-10 years I expect the same stuff to occur. Even when a good solution comes about, it will take time to move over. Look at businesses that held on to XP way after Windows 7 came out.

If I am wrong about how I feel, then so be it. There are others here with more insight I am sure. My boss calls SDN a solution to a problem that doesn't exist. Although, I can think of firewall scenarios that would be good to have automatically configured and I think the APIC idea works towards that. Just think about some of the stuff you do though and try to imagine that automatically happening, and for me it is hard to do.

Even just now, I was looking up SDN and it's still extremely vague and there's not a lot of examples out in the world. Programmers tend to be clueless about networking as well. A command line can't stay relevant forever and we're all likely to have to adapt at some point in the future, but who knows at this point. Most of the stuff I have read about relates to applications (which again, is the takeaway I got from the vague Cisco webex) and not how the rest of the stuff gets done. If it's possible for the programmer to automatically get his firewall rules put in or whatever, cool. I have never read about how SDN can turn up VPNs, figure out routes without deploying an IGP, turn up new interfaces for workstations, etc.

I had a total freakout at some point about SDN and tried to really discover the future of it and just couldn't get good information. Someone here has to know more about it (hopefully doesn't make me look foolish - but I did say take all this with a grain of salt!) so I would be interested to hear about it.

I should say that I am studying for my CCNP but afterwards might do an MS in IT management just to open additional doors since who knows what can happen. Think about this though. There have always been house maids, right? Then came the dishwasher. Then the automatic vacuum. Self cleaning ovens. Etc etc. House maids still exist though. Food for thought I guess. Or wishful thinking lol.

Good luck on your CCIE!

Bardlebee

I agree with everything you said. I will say I have very little view into SDN and the portions I've seen has been vague.

But lets say with the scenario right, that "Networks still need to be planned out". I agree, this factor alone means there will still be a need for network engineers. However, I am curious with stuff I'm reading (probably old tech) like PfR Cisco routing etc will put a lot of netops guys out of work. That is the guys who basically watch the network and if something breaks, then they fix it.

If SDN magically grew legs tomorrow, I believe perhaps a quarter (random fraction) of the people that were in network engineering positions strictly for break/fix scenarios may find themselves unemployed. That is the tin-foil hat theory anyway. Me at least just TRYING to get my CCIE has made me a better engineer and a planner, so I am hoping I would be immune to such things.

That being said I think the future of engineering is going to be more automated. The real question is, how much will that impact our employ ability and the job market. Will it be 5 percent of a decrease? Or 50? No one can really tell is my guess. If I don't get the CCIE in 2016 due to life (having a kid soon) I may diverge into looking into SDN while studying for it, just to be sure.

gorebrush

If you look at the way server technology has changed over the last ten years, then ultimately you can apply the same to networking, but virtual networking is way behind the proliferation that server virtualisation has enjoyed.

At the end of the day, in an SDN world or not, engineers will still need to know how IGP's work, so I wouldn't fear for my CCIE credential, or any other level for that matter.

Funnily enough I was browsing around YouTube last night and found a video by Aussie50 of some very old PIX firewall (and I'm talking OLD) and it was essentially a P2 server with loads of network cards loaded into it. At the end of the day firewalls and such forth are generally PC's anyway, so virtulisation of those will ultimately be the same as servers.

However, you still need MS engineers for Windows, so it's not like it's going to make network engineers disappear over night.

hurricane1091

Bardlebee wrote: »

I agree with everything you said. I will say I have very little view into SDN and the portions I've seen has been vague.

But lets say with the scenario right, that "Networks still need to be planned out". I agree, this factor alone means there will still be a need for network engineers. However, I am curious with stuff I'm reading (probably old tech) like PfR Cisco routing etc will put a lot of netops guys out of work. That is the guys who basically watch the network and if something breaks, then they fix it.

If SDN magically grew legs tomorrow, I believe perhaps a quarter (random fraction) of the people that were in network engineering positions strictly for break/fix scenarios may find themselves unemployed. That is the tin-foil hat theory anyway. Me at least just TRYING to get my CCIE has made me a better engineer and a planner, so I am hoping I would be immune to such things.

That being said I think the future of engineering is going to be more automated. The real question is, how much will that impact our employ ability and the job market. Will it be 5 percent of a decrease? Or 50? No one can really tell is my guess. If I don't get the CCIE in 2016 due to life (having a kid soon) I may diverge into looking into SDN while studying for it, just to be sure.

I mean, we actually have a lab for iWan here (I take no credit in it though, my boss and another guy have been working on it) and the purpose for it for us is to utilize backup circuits at branches that otherwise sit idle until the primary circuit fails.

Server guys had to set up physical servers in the past. Now they do it virtually. This actually didn't take away jobs, it created jobs if anything. If you can design and plan, you can probably be sure that your future will always be bright even if the technology changes.

Mow

Here's my take. I am not CCIE. I have kicked around the idea of it for a while, and even started down that path and quit. Now I am focused on another area, for a potential job I would love to get. I don't think getting your CCIE will be useless unless you plan to get it and then immediately quit learning forever. As a CCIE, I would expect you to see some really cool technology and have a better view of the overall trends of the industry, giving you ample time to learn what you need to learn to stay relevant. You will stay ahead of the rest of us simply out of the necessity to complete your customer projects. If you have some bleeding edge tech that a customer wants installed, who ya gonna call? CCIE!

gespenstern

Well...

If we think about a typical datacenter we'll find out that 10 years ago almost all the servers (at least Intel ones as IBM had virtualization decades ago) were physical, therefore they needed an RJ-45 patchcord put into their butts that connected them to switches.

These days I'd say >50% of OSes in a typical datacenter are virtual and guess what, they don't have a physical NIC and don't have to be connected to a physical switch. Yes, they have a virtual NIC instead and are connected to a virtual switch and this still needs to be served by some engineer, but there are two concerns here a) amount of service they need is arguably less and b) network engineers are pushed away from here by a virtualization crowd. Moreover, SDN resulting from all of this is arguably more flexible than physical networks and therefore is perceived as a way to go by the industry.

Will this trend continue in the future? I think for sure.

Bardlebee

gespenstern wrote: »

Well...

If we think about a typical datacenter we'll find out that 10 years ago almost all the servers (at least Intel ones as IBM had virtualization decades ago) were physical, therefore they needed an RJ-45 patchcord put into their butts that connected them to switches.

These days I'd say >50% of OSes in a typical datacenter are virtual and guess what, they don't have a physical NIC and don't have to be connected to a physical switch. Yes, they have a virtual NIC instead and are connected to a virtual switch and this still needs to be served by some engineer, but there are two concerns here a) amount of service they need is arguably less and b) network engineers are pushed away from here by a virtualization crowd. Moreover, SDN resulting from all of this is arguably more flexible than physical networks and therefore is perceived as a way to go by the industry.

Will this trend continue in the future? I think for sure.

Do you think it would be a worthwhile investment to learn and grow in virtualization then, say over traditional networking certs? Say I finish my CCIE, in theory. Perhaps I should pursue a VMWare certification and education?

networker050184

Automation is definitely the future. Whether that is ACI, or some other vendor or open technology, but the days of managing 100 devices individually by their CLI are coming to an end. Networking guys are light years behind the automation front that has taken place on the systems side. People are wanting to handle their infrastructure from a single pane of glass type or portal. That doesn't mean we still won't need to know the underlying technologies these things run on, but the way they are implemented is certainly changing. Regardless there will always be a place for smart people that stay up to date with technology. Things have already changed a lot in my career and I know people that have been left behind already. Working in technology is all about staying on top.

gespenstern

Bardlebee wrote: »

Do you think it would be a worthwhile investment to learn and grow in virtualization then, say over traditional networking certs? Say I finish my CCIE, in theory. Perhaps I should pursue a VMWare certification and education?

It surely won't hurt. I'm not sure about VMware's future though. But general virtualization, general SDN, general scripting & automation is certainly a way to go.

Bardlebee

Oh ok, that is good advice. Try to stay vendor agnostic. I just wasn't sure if VMWare certs were like the.... Cisco certs of networking. Where they hold more weight.

theodoxa

Bardlebee wrote: »

Do you think it would be a worthwhile investment to learn and grow in virtualization then, say over traditional networking certs? Say I finish my CCIE, in theory. Perhaps I should pursue a VMWare certification and education?

That is something I'm looking into. I learned some VMware (I had used Hyper-V and VirtualBox previously) setting up my CCIE Lab (VMware ESXi 6.0) and would like to pick up a VMware cert sometime. The VCP is not possible at this time because of the cost (VMware requires you to attend one of their bootcamps in order to take the exam) and because I'm busy studying other things (CCIE and Wireless), but they have something called VCA that is supposed to be a basic overview of virtualization and VMware and is a lot cheaper. I'll probably pick that cert up sometime in the near future.

I don't expect SDN to take our jobs as no matter what you still need someone who understands how the technologies work and work together.

networker050184

theodoxa wrote: »

I don't expect SDN to take our jobs as no matter what you still need someone who understands how the technologies work and work together.

SDN is just another way of getting packets from one place to another. That's our jobs as network professionals. We aren't here to work Cisco CLI, that's just the current way it's done. If you aren't looking at technology that way you just might get your job taken.

Bardlebee

networker050184 wrote: »

SDN is just another way of getting packets from one place to another. That's our jobs as network professionals. We aren't here to work Cisco CLI, that's just the current way it's done. If you aren't looking at technology that way you just might get your job taken.

This guys knows whats up. I agree.

linuxabuser

theodoxa wrote: »

That is something I'm looking into. I learned some VMware (I had used Hyper-V and VirtualBox previously) setting up my CCIE Lab (VMware ESXi 6.0) and would like to pick up a VMware cert sometime. The VCP is not possible at this time because of the cost (VMware requires you to attend one of their bootcamps in order to take the exam) and because I'm busy studying other things (CCIE and Wireless), but they have something called VCA that is supposed to be a basic overview of virtualization and VMware and is a lot cheaper. I'll probably pick that cert up sometime in the near future.

I don't expect SDN to take our jobs as no matter what you still need someone who understands how the technologies work and work together.

You must have missed that 3 year long saga that is Stanly CC.

Dieg0M

SDN is just another attempt in centralizing the network by using software. Centralizing networks has been attempted in the past and has failed as it is technically impossible for a distributed computer system to be consistent, available and partition tolerant at the same time(as per Brewer's theorem). In my opinion the best use case for SDN will be to offer a hybrid system where we centralize some of the network policies and we distribute network routing and switching control-planes.

This said, programming is a useful skill in general in IT so you should still look into learning a scripting language after or during your studies.

joelsfood

Being able to script/centrally control something, whether it's network config in SDN or system configs/rollouts in vmware/aws/whatever, doesn't help if you don't understand what you are scripting/controlling/creating. CCIE isn't going anywhere. Learn your networking, or someone better will take your job.

Danielh22185

I attended a Cisco Users Group meeting recently at the Cisco Richardson campus here in DFW. The topic was SDN and APIC. It was all very interesting. It also kind of made me freak out a bit as I am only about 3 years into my networking career and have CCIE on the future horizons too.

My take from it is... it's still quite a ways off for being mainstream. However it does look to have some definite advantages. I think where it will shine is in its early stages on the deployment and operational fronts. Deployment meaning you can write up a configuration for a device by means of a script to deploy to the device and turn it up (this really already exists today but how it is managed will change with SDN). Then offer day-to-day monitoring / operational maintenance via the a centralized interface.

However this does not actually put intelligence into the software either. You will still need an expert of the infrastructure from a networking technology standpoint whom can implement the network elements into the software that would allow it run transparent to the users. Subnets, traffic routing, filtering, etc. still need to be planned and implemented accordingly.

Also there will always be physical. Even if you are maintaining one giant central computer you will still need on-site personnel to maintain the equipment. I personally don’t see it becoming something like what we see in Sci-Fi where it literally just looks like one super hive mind computer with neato flashing lights, emanating from a giant hunk of metal. Yet, I think it will slim down the physical to an extent where it is feasibly possible.

To be honest I see this marketing concept or a tool for a small company that can centrally integrate itself into one system because the small scale of their network infrastructure. However for super enterprise networks or even larger (the ISP world) the change will take a lot longer (A LOT LONGER).

Will SDN potentially eliminate jobs? Sure, as the need for operations roles will shrink to only what is needed (we do that now); but I think a reverberation of that will be a creation of new jobs specific to the new technology.

I like to think of this as the scare we had with IPv4 depletion. We had to develop a new technology to maintain what we had while the fix (IPv6) became more mainstream (It still isn't mainstream). This idea was when something actually was breaking down and needed more immediate action. Technically as we look at our current infrastructure nothing is “broken” per say… So, we have time ahead of us to plan how SDN will integrate with how we operate today. “If it ain’t broken don’t fix it!”

I think so long as we as network engineers continue to embrace change rather than rejecting it we will easily be able to maintain our jobs and adapt to how they change. I can speak from experience that I know some people that rejected change to networking technology earlier in their careers or in the present and I now am their escalation point for network troubleshooting So in the now, we all should continue to learn as much as we can about existing network technology so we know how in the future it may change and we can adapt according to that change.

hurricane1091

This relates I feel like.

I've sat here and made an entire DMVPN lab because our environment does not work. My lab works, and I think I know how to fix it in production. Among a few things, we're going to need to add all the physical IPs of all branch offices to the inbound ACL. Manually, this would be annoying. My plan? Write a Cat Tools job and send it. Which works for us because the ACL is named the same on all branch routers.

What would SDN possibly be able to do in the future? Probably the same thing for me. Except a little easier probably. Maybe somehow without me pulling the IPs manually too. It could be helpful, but I still needed to know to go do it, why to do it, etc.

That's my take on things.

Priston

Reinventing the Internet #WeAreCisco - Cisco

https://en.wikipedia.org/wiki/Content_centric_networking

https://en.wikipedia.org/wiki/Information-centric_networking

https://en.wikipedia.org/wiki/Named_data_networking

Segovia

Just a little input on this SDN topic. At my org they do have SDN pane of glass type of solution set up through solar winds. This enables the lower tiered IT guys to troubleshoot switchport problems and selects from a few different pre made scripts. In my short experience with this setup, it was mostly implemented to take the load off of network engineers. For every IP phone or workstation that needs to be changed or moved, the tickets would be escalated higher up. Now the desktop support takes care of these simple tasks. I met one of the network engineers, he spends most of his time whiteboarding and labbing and designing other solutions.

For what it's worth, SDN won't make your CCIE worthless. You won't be making scripts on SDN that will totally change a routing table or ACL for instance. Higher level networking problems require higher level thinking, and will never be point and click.

theodoxa

linuxabuser wrote: »

You must have missed that 3 year long saga that is Stanly CC.

I'm on their wait list. No idea how long that takes, though or if they're even still doing it as the "Upcoming Classes" listed on their page were 7-9 months ago.

shednik

SDN is nothing to fear you just need to be open to new ways of doing things. SDN will just be an overlay that will migrate *some* of the existing features used today on a router or switch and put them in at a different layer. No matter what you will still need a network to connect everything together.