previously but like if i'm starting today is there any kind of path that you would recommend someone take based on what you've seen like do i become a ccna do i go and do devnet is it what are the sort of the technologies we did this before but i'd like to get an updated list are there any top five top 10 technologies that you think are really important and are there any is there any path to get there well you see it depends on whether you're looking short term or long term short term today i would go with automation or cloud these are the things that will be probably in greatest demand in the next three to five years long term learn the fundamentals ip routing hasn't changed in 40 years ethernet hasn't changed in 40 years everything we did was up the speeds 400 gig ethernet is still ethernet yeah yeah it's still running ipv4 and ipv6 on top of that and it's still transporting more ipv4 than ipv6 tcp is slowly changing with quick and things like that but ospf is there bgp will not go anywhere arp will be there forever because you know if you have layer 2 and layer 3 you have to map them somehow so regardless of what you know pays the bills learn the fundamentals because eventually you will need them yeah and eventually every environment will get to a point where the network will be down and who will troubleshoot that the python guys the cloud guys or someone who took time and learned the fundamentals [Music] okay so what are the what are the skills that you reckon are important and when you say fundamentals how would i learn that ccna is not bad yeah but look at ccna not as assert well yeah we said it's you know just to prevent being thrown out of the huge pile of cvs yeah just go for it and get the cert and you learn something but more importantly figure out what are the things that you should get out of ccna how does ethernet work how does routing work what is ips subnetting do they do poor channels yes poor channels spanning tree ospf that stuff and then you know what build a lab with cumulus vx boxes and do that same thing on linux just for the giggles and then you'll learn how yeah it's different cli it's different approach it's monolith versus multiple processes but in the end you'll get ospf running between a cisco box and linux host and then you will understand how things work a little bit better than before so when we spoke quite a what's about a year ago already you mentioned like some skills like you need to learn linux you've kind of mentioned that again um so for the short term so i mean everything short term these days but let's say in the next two to three years what or five years even what do you think are the top skills that i should learn if i'm trying to get into this industry you've mentioned ccna kind of like gets you the basics of networking um but do you have any other like sort of high-level skills that you think are really important like if you had to choose five things or ten things what would you pick linux perhaps and then well uh we mentioned learning right yeah deep work i i read the book deep work by cal newport i think yes it's an awesome book and it sent me thinking so you know your brain is your most valuable muscle and if people are willing to spend time in the gym and train the other muscles you should train your brain as well which means that you should be able to you know take a problem and focus on that problem and work on that problem for like three to four hours uninterrupted and if you can't do that you have a serious problem the other thing is just you know do honest work in the evening look yourself in the mirror and ask yourself did i do everything i could do today to get the job done and if the answer is nah then you have a problem i'm going to push you ivan because you've mentioned like it's more philosophical type stuff which i think is really really important but are there any practical skills that you think i should like you mentioned cloud should i like and i i don't if you're not happy with the question that's fine but like you mentioned ccna should i go and do aws like associate cert what what do you think are the big trends or the certs or the knowledge that i should try and get in the next few years well uh definitely so don't ask me about the search because honestly i don't care about them yeah that's fine so give us like sort of the technologies yeah uh i told you uh you you need some basic networking fundamentals yeah so how do networks work then you know cloud work clouds work slightly different than the traditional networks so learn how cloud networking works don't only focus on networking because you know once you get to the clouds it's all intermixed yep it's vms it's storage it's containers it's lambdas it's networking it's load balancing it's uh security packet filters application firewalls and you know what it's all free all three big cloud providers have a free tier use it the only thing it's costing you is your time and yeah then the next thing obviously is figure out how to automate stuff so for example you're studying for ccna for god's sake don't use gui to set up your lab [Laughter] use ansible use cli use api use anything but gui yep you want to you know set up your ospf lab don't even think about logging into five routers and typing in ip addresses what have you learned typing skills hooray we really need that that learn ansible learn how to build configs push configs to the boxes you'll master ccna you'll master networking you'll figure out how things work plus you'll learn something else once you start doing that don't have script.1.1.a.3.zat on your folder use git branches figure out how you can experiment and test things out and you know destroy failed experiments and accept successful experiments with for example git branches everything you do there is always something extra you can do to learn something else not just the one thing that you are working on and you know what it's all free git is free ansible is free linux is free baseline cloud here is free cumulus vx is free juniper vsrx 3.0 is free arista eos is free well the vagrant box yeah cisco is not free who cares oh yeah nexus 9000 v is ios is not free forget ios let's move on sorry cisco get your house in order that's funny that you mentioned that because one of the questions i wanted to ask you ivan is cisco juniper arista what should i learn i mean you've mentioned cumulus quite a few times so you'd all of them all of them it doesn't matter you know it really depends on do you want to be a cli jockey or do you want to be an engineer yeah do you have a brass mechanical engineer and an iron mechanical engineer and a steel mechanical engineer i don't think so good point so why do we have a cisco cli jockey and a juniper cli jockey and arista cli jockey they all run ospf it's just configured differently well now arista is using industry standard cli so in other words you're saying learn the technology like ospf don't worry so much about the cli is that what you're saying yeah no uh honestly what i did was uh probably two months ago now i said well now i want to build a tool to set up my automation labs so that yeah yeah so i built that python tool that takes the lab topology and creates device data models and now i can do initial config and the next step i'll do for example ospf or bgp or what have you but as part of that experience i built boxes for as many different platforms as i could get my hands on nexus os is downloadable vsrx30 is downloadable yeah uh vmx i think isn't but it's a mess to set up anyway because they want to have two vms one for data playing well for control planes so just go with vsrx yeah arista is downloadable um nexus one 9000v is downloadable cumulus is downloadable the problem i had was that i wanted to do this on kvm because you know learn a new stuff vagrant on mac os or windows it works out of the box you can just pull down the boxes it all works it's all crap but it works and you learn nothing new so i built all those boxes myself and now i have a running uh liberty environment on linux box with automatic provisioning and i just built the laptopology and bam it goes up and now i can really test anything i want between like four different vendors that's totally different than trying to learn you know the cli commands for a single vendor and it's not harder because they are similar well ignoring juniper juniper is special but even cumulus you know once you get it up and running it's uh yeah layer two stuff is special interfaces are special but once you get to ospf and bgp it's free range routing which is quagga which is really cisco syntax ivan what do what do you think are the next big trends what do you think is a good trend or trends to look at if you're just starting in this industry so are there any waves that i can ride if i've just started for a young person going into you know i.t i would say focus on the cloud okay because everyone will have to do something with the clouds eventually and traditional environments will have no idea whatsoever how to do that so you know it's like when we went from cobol to javascript 30 years ago or whatever it was or or was it java probably more java than javascript uh there were tons of people familiar with the old technologies and no one familiar with the new ones and everyone wanted to be hip and use the new stuff and probably cloud is the new stuff at least in some geographies obviously not for silicon valley obviously not for startups but for traditional i.t do you think they can spare a cloud something up there in the sky isn't it yeah exactly it rains right yeah so ivan you've mentioned get already so it's interesting what's your opinion of cisco devnet and that those kind of certifications they can't hurt yeah so you see it's always the question of where you're coming from and why you need certain things yeah so if you are an existing networking engineer and you think that you have to become a programmer which is a wrong idea but we can go there some other time uh then yes something like devnet is the right thing to do because you already have networking expertise you know nothing about python and rest api and git and all that stuff you have to learn those things if you're a developer then you already know all those things and if you want to be you know a network automation developer then maybe you should focus more on understanding how networks really work because you know you have the developers that can develop anything and are not good at anything and then you have people who have actually worked in some industry for ages so for example you might have a mediocre developer who has been working in chemical industry for the last 20 years he will beat any wizkid because he knows all the dirty details all the requirements all the regulations all the audits you have to go through whereas you know someone who might be a brilliant programmer has no idea about the industry and the same thing if you want to be a good developer in network automation space then maybe it's more than python maybe you should know how networks really work do you do you think there's going to be a trend where network engineers the traditional network engineers are replaced with guys who do automation is it are we kind of all forced to do automation now well you see uh the way we've been doing things in the past is crazy yeah i mean just configuring the same crap on thousand remote office routers manually and using excel to replace the ip prefixes in your configuration it's the future what stone age are we living in exactly uh so yeah we we need to clean up the crap that we're in and one way of cleaning it up is through network automation you've been in this game for a long time what would you suggest someone do if they're starting out so in other words you know if you could go back in time or let's say you were 18 years old and you started today what would you tell your younger self oh my younger self okay or anyone in the industry you know what would you do that's probably a bit different but yeah let's start with my younger self okay the most important lesson technology doesn't matter as much as you think it does so honestly you know when you're young and brave and stupid you think that you can solve all the problems with throwing technology at them doesn't work that way then you play macgyver and you solve impossible problems by configuring stuff in the way it was never designed to be configured it works in a lab it breaks in production you are caught at 2 am on a sunday morning you have to rush off to the customer to fix your crap you learn then you know sometimes it might work for a year or two and then you install a new software release and of course it crashes because no one ever used the same box in the same combo with the same config and you felt so smart and now you feel utterly stupid and mad at the bloody vendor for sending you the bug forgetting that you are the only idiot in the world using that particular code path so minor things like that oh there's one more uh old people aren't stupid uh contrary to what all the youngsters believe and i'm not like defending myself i was in that position uh if they do something and you think it's totally stupid it might be i mean after all they might have stopped thinking after a while and just do things the way they do things because they've always done things that way but there usually is a reason behind that the reason might be irrelevant or it might be wrong or you know whatever but you have to find out the reason first before you're trying to fix stuff uh there's this english saying and i always forget what it is about not moving offense that you encounter if you have no idea why the fence is there yeah so i mean basically you were like that when you were younger i think most of us were oh yeah of course you knew better than other people everyone else of course yes especially if they were the smartest person in the universe of course so i mean you know the running joke today is like boomers get out the way so you you you're basically telling yourself if you if you were younger that's a mistake is that right well think first that's the most important message there's a reason people are doing things figure out what the reason is don't just rush in and you know take over and do your stuff thinking you know best sometimes you do know better sometimes they do stuff for stupid reasons sometimes they really are boomers but sometimes there's something behind it that you just don't get so i mean if my counter to you would be like technology's changing the world i mean when we were kids you know facebook didn't exist instagram all these you know massive websites so surely technology is really really important or what are you saying to your younger self that you should learn apart from technology uh listening that's a good one people skills communicating listening did i say listening you did listening uh the the most important thing is listening to what people are telling you and trying to understand what they're really saying not what you're hearing because sometimes a lot of times we have a problem because they're using their terminology and you think it means something else and in the end you do something or you implement something that is total crap like my favorite one you know it's stretching vlans across two data centers layer two your favorite layer two layer two into the cloud yeah that sounds like a great idea yeah it's awesome i mean earth is flat right what could go wrong um i've heard this a lot people talk about people skills and doing things differently but the question i wanted to really ask you is give us an example of where you really messed up where if you if you had your wisdom and the experience that you have today you wouldn't have perhaps gone and done something so it's always nice to get a story of where you know you made a mistake uh like uh things like debug ipbgp on the core router that connected the uh a country to the wider internet that sounds like a good example so tell us a bit about that and well you know we were setting up a bgp feed with the upstream provider and we already had a number of production customers on that box that was the router that was the hub of the commercial internet in slovenia and bg we were getting some prefixes and i got the route maps wrong or something so of course what do you do debug ipbgp let's see all the updates well yeah 100 000 updates later on a console port of uh cisco router the real problem is for the youngsters who don't know that in the good old days i don't know whether that still works today you were able to break into ios on the console board so you would send the break signal and you would be in a debugger yeah which means that console must be the highest priority interrupt on the system otherwise you know something could lock it out yeah now imagine that you are generating thousands of highest priority interrupts per second just because someone is printing debugging messages to the console obviously the router just goes like ah i don't want to leave anymore you know it's great to hear stories like that because you very respected and i mean these days people look at you and think you know this guy can't make a mistake so it's great to hear that no you know actually designing a network with 20 parallel eigrp processes you did that yeah it worked yeah but is that an example of you like like you said earlier you taking your technical knowledge um and trying to fix a bad design like i mean i think you've mentioned this in other in other videos where you talk about like ospf craziness and just trying to sort out mad stuff because you you can yeah you know when you're young and crazy you think that you can solve every problem out there with technology yeah and yeah my my eigrp design was one of those things effectively the customer needed mplsvpn uh and we told them how to do stuff install core routers and then install route filters on the cover router so that the edge routers will only get certain prefixes they had you know shared address space for everyone and the mpls vpn wasn't even on the drawing board so we couldn't use that and the customer said no no you know for political reasons it has to be frame relay in the middle from company x and routers at the edges bought from our company you know the pie chart diagram every supplier has to get certain percentage of the business so no one will complain layer 9 or 10 right yep in the osi model and so we had to do something and they wanted i think they wanted to implement the whole thing with frame relay dlcis so that you know every tenant would have its own partial mesh of frame relay circuits but there was this still this problem of the well there were two problems one was that they had the central site and everyone had to connect to that central site and then you get route leaking and isolation is gone and even more interestingly they had multiple tenants sitting in the same building and they were not keen on buying a router per tenant i don't know why i mean after all they wanted to have frame relay and you know anyway so we solved that thingy by running yeah one er grp process per customer man yeah and i even asked the friends at cisco tech what they think about that and they go like yeah it should work but do remember that you are the only one in the world doing that so you might encounter a few extra bugs oh wow and i'm assuming that's what happened yeah no actually that thing worked it was too successful so you can like solve every problem now uh well yeah they thought so so what happened was that the customer you know took that ran with it and then they figured out that they needed to provide internet access to those locations and central services and something else and oh and they had regional hubs so in the end it was this orthogonal matrix of ergrp processes per regions and erg processes per tenants and two-way redistribution between them but at least you had job job security didn't you well someone another [Music] consulting company took over that design and they were milking the customer for years i can imagine but yeah they had job security you
2021-03-27