Some boring network engineering interview questions and how to replace them with smarter ones

hi my name is cam and it's a very interesting topic today we're gonna talk about some boring network interview questions and that's a very interesting title by the way I introduced myself to someone right outside and they called me oh so you were the boring guy yes it's a very interesting topic let's get it started and talk about this thirty minutes is a very very short time to cover this topic however I'm still gonna start in the story the story is many years ago I was on this phone screen I was actually being interviewed with a major company in the Bay Area and I was not talking to an IC I was talking to one of their leaders gentlemen asked me a very interesting question the question was what's the total number of passengers they pass through the SFO San Francisco Airport every year well what he did not know about me was on a professional traveler or travel a lot I've been to many major cities and almost every single major airport in the country and I talk to people a lot about their travel habits to travel patterns so took me a couple of minutes to come up with the process crunch a bunch of numbers I got back to him with a pretty good estimate an estimate that later I found out was pretty close actually amazingly close to the actual number but only on you it's a variance a very big neurone to share a secret but anyways I was calling expecting a trophy but he abruptly ended the interview the phone screen and I got the rejection email whether it was cheating or something else but they were not interested in continuing the conversation because my estimate was way too accurate well you just talked to Ron Meyer I guess but that was the case that was around the same time I started thinking well maybe there is a way to restructure Network interviews better be smarter and many cover areas and skill sets that are not necessarily things that people can memorize and when they answer that you get really really surprised so that and also defining the bar so where is the bar for a junior level person for senior level person for a very senior level person that's how this whole thing got started I work for the University of Southern California my teams are responsible for anything network related basically we start from the fiber infrastructure we have across the city of Los Angeles all the way to VX LAN and BGP and OSPF and you name it wired and wireless very proud of a team this is the topic today we're going to cover here by the way the previous session was great this is a more technical session I guess complementing a session so let's get started and have a look the way they structure this interview for the next 30 minutes we're going to talk about a traditional network engineer and what it means when we talk about someone who is a traditional network engineer Muslim you're referring to you someone with routing and searching skillsets also we have a lot of areas that over time put together as true areas that we as a community traditionally tend to cover technically when we are interviewing people and also I do have a bunch of questions here just samples or to show you what are the questions that we ask and maybe a couple of different ways to ask the exact same question but in a different way so let's have a look and say so from a traditional perspective when you're interviewing a network engineer are you expecting the person to know tcp/ip or UDP IP I guess this is the base right and then we talk about something called OTP of choice the person depending on your network should know OSPF or is is maybe more maybe both and BGP is a very interesting one we do have some slides on BGP as well we cover BGP and that actually changed the has changed over the past 10 years a lot through other areas that you can call them areas of expertise the service provider data center wireless security advice this particular presentation given the time that we have we will be focus mostly on traditional running and sewage engineers but you can apply pretty much the same concept to other areas so for whoever is interested in literature we have great expectations these are the things that we again traditionally tend to cover when you're hiring a network engineer when you're interviewing a network engineer these are the topics that we ask we want the candidate to know the basics of tcp/ip basically the structure of the packets maybe some offsets maybe the sequence number maybe some of the flags we also talk about the session establishment process the congestion control mechanism is a very interesting one let me give you an example when you're interviewing someone a junior level candidates something that you cover let's talk about three-way handshake when he knows the curve a handshake may be a little bit slow start and what happens when you get back a loss and then later maybe more senior level candidates the behavior after the packet loss how far we go down we go down 50 percent all the way down or maybe to one segment and then back up for the senior level candidates depending on what you really need in your particular network you even might want to get into things or different implementations we can go as far back as Tahoe and Reno start the story from there and get to Cuban cubic and then see how they use those in your network also we have the ojp of choice again OSPF an RS is two interesting protocols right we talk about the messaging system we tend to ask questions about maybe some multi level design LSI's and LSP is different coins again another interesting example here interviewing a junior level candidates when you're talking about OSPF let's talk about LSA 27 okay now sort of you were trying to find out whether the candidate knows NSSI and how it works a little bit more senior you talking about the translation process as the LSA hits IPR a little bit more senior I have multiple ABR's how does that translation process works now you're talking about very senior we're senior-level candidate and you ask questions like okay how do I minimize the number of LSA fours in my network so there are a lot of opportunities here and we have very interesting examples on each one of these BGP is a very interesting one so four major areas that we again tend to cover one of them is BGP attributes we all want to talk about different BGP attributes and then there is a decision-making process there is no way that you walk in a network engineering interview and you do not talk about the tributes or the decision making process then there is traffic manipulation inbound and outbound it's so so famous that I can help you memorize and pass that question in five seconds someone asks an inbound question AAS path remedy someone else came out that question it's either way to local pref and then we have be cheap in other areas keep in mind BGP has changed again over the past ten years we used to add vgp just as an edge technology between your network and the rest of the world that paradigm has changed altogether now have BGP my data centers we're gonna have a reference point presentation for us coming up and interesting again same thing we have in enterprises may large enterprises today they use BGP so there are some great opportunities there too but there is a problem with that all of these topics that we tend to cover they go after technical depth the technical expertise of the protocol candidates how much do you know about this ok I'm checking all these boxes and that's how many of these boring interview questions were actually created there is one big problem with that approach and that's a leveling thing I have heard and been to many interviews that we talked to a candidate at the end of the interview ok there are conversations whether this candidate is a level 8 or 11 9 or maybe a level 7 ok it's 11 9 what's on not hiring him he's just too senior for the opposition we just do not have the right data to make that decision again another reason those traditional questions cannot answer questions like this cannot help cases like that and especially if you have two candidates same level in a video two candidates is Friday afternoon were making a decision which one was better well they both knew they different LSPs TLD talks or is is they both names is is in te may be the LS different all different kinds of other says no SPF even maybe LSI type new LSA is that we use for ipv6 but they both did which one was a better candidate another problem with that today we are interested in three great qualities and the previous panel kind of covered that when I was creativity innovation and passion you all know creativity and innovation these are the two characteristics that I don't think we need to get into a lot of details we have pieces today no matter how junior you are we have pieces of code that people need to put together and create something else same concept applies to many other things including tools different parts of the network so it's sometimes about building something new using things that people already built what passion is another thing it's a very amazing one sometimes I interview candidates you'll see the passion the passion is definitely there he or she is passionate about the brand about the company not necessarily this particular technology this thing that is broken and has to be pitch for the customer that is kind of passion we're looking for I'm looking for that engineer chases me down the hall says look we talked about that solution but keep in mind are two other ways to go around this those who may not be necessarily technical but here or those two and these are the lot cost and different timeline comparison all the items about them so that is the passion and again boom battle appears fight that kind of questions and just being hundred percent focused on those technical areas that we all traditionally covered is not going to help you answer these questions and finest qualities in your candidates so we're going to explore some of those topics and it's it's kind of amazing whenever our presenters I have a much larger version of this deck always every single person I never asked any of these questions but I've heard them a lot anyways let's have a look at the questions I'm sure some of these questions are pretty familiar this is a very I guess famous one differences in TCP and UDP let's get let's go to some details here when we ask this question when we ask this question intention is we want to sort of see whether the candidate knows about layer four right that's what it is the problem with that is our ears a brain as interviewers we are waiting to hear a word and that word is reliability if this thing if the candidates uses the word reliability this thing is reliable or act back you you black of a choice you're not going to press the candidate in harder that's okay he knows it let's check this box I move on to the next question we have seen it many many times and we covered that and it doesn't really give you any details actually leveling creativity okay whether this candidate is gonna be able to create anything for me anything meaningful if he or she is put in a situation this is a scenario that we came up with it's a very interesting case a very simple case I'm not saying it's technically 100% accurate from an aviation perspective but that's a great example you are now hired as an architect as an engineer for an airliner and you're trying to make this decision they have a lot of plans of an air just trying to locate those planes every five minutes every five minutes means 8090 miles right kilometer is probably so what is the smartest way you have to make a choice TCP or UDP and design the system for the company so when you ask a question like this you were trying you were you were engaging you were asking the candidates to start a conversation with you rather than just a yes-or-no answer or just a technical keyword the candidate needs to understand the stakes all the risks we're talking about people's lives and also the formation of and look at this you have two choices you have TCP and UDP right or TCP and UDP well down the tcp path you're gonna have reliability meaning you're gonna send that message you're gonna make sure the message has arrived with UDP you're not gonna wait for anything you don't have to establish the session the round-trip top time which potentially could be pretty long we're talking about hundreds of miles that's not a factor there anymore but problem with UDP is if you lose one of these handshakes one of these connections back to your central points or your lesson or in general you're gonna have some problems then your estimate might be off but 80 90 kilometers so the candidate can just take sides and you can engage in a very productive conversation now you have a chance to defend your reasoning to give me enough reasons to convince me this is this is the better choice and here are the reasons this question we did not ask anything about the cost about the time line but smarter candidates they take it one step further they're gonna tell you that look this is good on my choices A or B however keep in mind we there are other technologies after or this is the number of lessons you need or maybe you don't want to forget about this let's build something else on top of that and sort of make up for this technology which may not be reliable so that's a good example I think from candidate engagement perspective and you can exactly see the thought process they migrate candidates always ask a lot of questions they get a lot of details come up with something more than just a TCP or UDP answer that you check a box and move on this is another one different setting always CF&I Esaias I have no idea how many times I've been asked this question I've heard that question so the reason we asked this question as usual we want to know whether the candidate has any exposure to these protocols I mean what else right there are tables after their books and tables after you can just look it up and the table always have on one side is on one side memorize it and pass the interview easy but the bigger challenge there is these two protocols under the microscope they are not very similar yes they are running protocols yes they are link state yes there are these linker states and an algorithm conversations both it pretty much it stops there everything else is different the messaging system is different neighbors different different modes are different how you design a multi-level network is different so these things are all different but also I don't see a lot of room for creativity there is a bigger risk if this kind of question that I want to take advantage of this question and sort of expose that let's look at the next slide it's called bias interview let me give you an example and it's not really directly related to this particular question but it is a big challenge in our space some one gentleman rock here I asked a question can you describe this room well okay we can talk about the square footage we can talk about the ceiling we can talk about the roof we can talk about the building itself we can talk about this stage of course we have a lot of things here but as you name all these things about this room as you're trying to describe this room there is a problem with that the interviewer is just shaking his hand his head like this not that one you're close but not that one not that one either so now you've gone through ten different things about this room but you have not yet yet that very thing that he has in mind okay then what is it that's about the podium before God talking about this yeah and that's very important and this is the interview session we go to the debrief I'm gonna go to the beer debrief it's called hand holding right I'm trying to explain to my colleagues how the interview went he needed a lot of hand-holding I ended up giving him a lot of hands for him to start talking about the podium he covered everything even even the start even the sky talked about the carpets but I forgot the podium this is like an important right so that's biased interviewing questions like the difference between OSPF and is is like you can list long little items that's not necessarily something that you can ask and say to say it's a safe and fair interview here is the scenario I used to well basically replace the other one right so I have very limited amount of time it's a research being very limited amount of time and as usual don't have enough money right now the question is I'm working on a new version of IP let's call that IP v6 point five and I want to come up with a routing protocol able to support IPs six point five so what will be your choice always BFRs is probably no right or wrong answer but this is exactly the problem that this industry something from the room I'm sure we solved at some point maybe twenty years ago right this is a real-world question we had these two protocols and other routing protocols the question was ipv6 now is of reality how do we make our running protocols compatibles able to support those things when you ask the question like this the candidate needs to understand limitations the time is critical so I'm not gonna design a running protocol from scratch always be a v3 was running protocol from scratch great masterpiece but something from scratch we use the idea but no one used v2 code to make changes story to add component that and also fewest number of changes that's a goal you want a candidate you want to see your candidate able to articulate their reasoning I pick eius eius because it's based on tlvs till these give the protocol extendibility my protocol is extendable I don't have enough time so I'm not going to design the routing protocol from scratch but a better candidates may be a great candidate would give you the ISO sensor but at the same time toward the end of the answer is gonna say by the way what makes you think you should use Isis or SPF how about BGP how about extending BGP we use b2b for a lot of things you can update analyze how about we use that for that purpose this is great this is a point Oh awesome absolutely so here we have three BGP scenarios most of them very famous one I'm sure you've seen all of them BGP decision-making process candidate knows this or not we start from next hop availability and wait go all the way down it's a memory test question trust me and I have seen people flash cards everyone's memorizing this case number two you've heard of different ways that we use to describe attributes well-known optional transitive non-transitive again another area that's a memory test candidates do memorize these things you're not gonna get anything out of that interview if you ask a question okay Emma do it is transitive is not transitive okay what what does that really mean he knows that okay good memory I guess and this is a great replacement again the same scenario same framework you're running out of time you don't have enough resources to design something from scratch but at the same time now you're developing a new experimental version of PGP what would be the very first or maybe couple of first few first attributes that you have to design keep in mind your router should be able to talk should be able to send messages to neighbors and process at a very minimum that bare bare minimum process and support BGP running protocol and this is one of the questions that always suggest to replace in this case your candidate they need to know well none a mandatory and how these things work what does that really mean how about my BGP knowledge is it just limited to simple laps I brought up a bunch of BGP different aliases or understand behind the saying what each one of those keywords really really mean and how we can take advantage when we are designing networks Noma 3 this is an interesting one influence inbound traffic or traffic it's one of those famous questions that people ask all the time and of course it shows you better whether the candidate has done this at least once in the past or maybe they memorized that here is a problem with that again creative no creativity there no leveling there and that memory problem this is a better replacement listening call this for a second now you have this junior engineer and that your junior engineer is making a great suggestion says we want to influence our inbound traffic let's use the MIDI you are the senior person now you are in a position to challenge that decision how would you challenge that decision what's wrong with using the meed if you're trying to influence your inbound Internet traffic I'm not talking about to a SS I'm talking about Internet traffic when you ask a question like this you give the candidate opportunity to expose their thought process now use now you hop and see the limitations how definite how does the network now you will get down to the two key words who transitive and non transitive Internet is a massive space with thousands of aliases can I use something that is non transitive oh maybe I should use something that is transitive maybe a s path is a better option this is this is one of those bonus questions one of those questions that toward the end of the interview with someone nailed that answer I want to use a staff great perfect let's use a yes path this is a bonus question this is something that I think we can add to our question sets and ask okay but there's a player problem with that is path is correct but I do not have enough resources I do not have enough skill set some people to make any immediately changes come up with something else other than vgp missing the BGP attributes now here is where you can see the key or candidate understands longer prefixes or shorter prefixes and how they're rather across the Internet that's one way of doing it smaller candidates might say getting asked the question again that's a question again he or she does not hear the word available or availability availability is not a concern so who cares let's stop advertising that particular prefix out of one of your circuits I know I mean if you were to lose the third circuit you would definitely disconnect it if you were that prefix but that would still satisfy this question you're not gonna have availability but that's one of the possible answers as you see you're all sort of different ways to engage candidates to start conversations to see the thought process to see if they can go beyond what the ebooks is these are really these are great skill sets that we want to see in our candidates creativity imagination what's a passion piece of that the passion is exactly that right I want to do this but what if the engineer is not available my BGP God is not available let's do it this way let's make some changes your traffic's prefixes in terms of the length and solve that problem interesting areas I am done with this presentation and we have four minutes left the whole message before end this and open questions the whole message here is there are many many smaller ways that we can engage our candidates and have a conversation with them than just asking simple questions textbook questions that have text which answers there is nothing wrong with asking textbook questions but at some point you want to see the thought process you want to see how they can be creative if not every single two is available to them do they care about the cost do they care about the scale do they care about things like my timeline my customers stakes how important this thing is what if I make a mistake the previous panel we talked about risks whole key point here I think they all mentioned is calculated risk is this a calculated risk or not if something is not a calculated risk you cannot calculate the risk we're not going to take that but you have to see that thought process in your candidates when your question is difference of the United States and I will speed up you're not gonna see their ability to go through the risks and assess the risk and do some calculated risk assessments so that was the whole point I really appreciate it thank you so much I don't mind questions please hi I'm Chris Wood build with Salesforce I wanted to know if I noticed that the example questions that you threw out I realized these are all examples but when you do these kinds of interviews and this may have changed over time this might not been the kind of question you would ask maybe 10 years ago but you definitely should people definitely should be asking these kinds of questions now about automation and orchestration and software skills is that something that that you touch on in these types of interviews as well absolutely that's a great point I do not have the automation of slides here but automation keep in mind automation we went through a long journey we started very very basic things then leaders who screen scraping and just basics their lives a great candidate if any kind of automation skill set modern automation skills that can answer my biggest concern my biggest concern is over trying we purchased five different vendors now I have all sorts of codes but here is my goal here's my intention I want to have one configuration I won't have one operation model how do we do that how do we achieve it it helps you see whether this candidate cares about the latest developments in this industrial previous knowledge I think maybe even this wants you we had conversations and things like open config like yang models all those things these are the things these are the qualities you want to see in our candidates so absolutely yes and the solutions that you're designed from a design perspective is this something that can be automated some solutions cannot easily be automated that's another key point and simple better network design and architecture there might be five six waste as she's the same goal if simplicity is possible then probably yeah that's a preferred option if not then what's the right way to automate that solution great question thanks any other questions all right thank you so much everyone for your time I will be happy to answer your questions through email LinkedIn please feel free to reach out I would love to have these conversations this is something that I as an individual or my team in my group we cannot do just all by ourselves we want to hear you want to hear from you I would love to hear from you your experiences if Google interviews about interviews and all the friend aspects gaps that use interviews things that we can do and set the right bar for this software engineering they've come a long way software engineering now has a very established interview process network engineering we're those guys fifteen years ago we have to catch up I'd be more than happy to get in touch with whoever in this room is interested LinkedIn or email please reach out we can have a great conversation thank you [Applause] you

Channel: NANOG

Views: 7,248

Rating: 4.8032789 out of 5

Keywords:

Id: g35UumfZ-H4

Length: 30min 25sec (1825 seconds)

Published: Mon Nov 04 2019