Kentik - Network Observability
More episodes
Network AF  |  Season 1 - Episode 6  |  November 23, 2021

Mentorship and shared languages of network engineering with Cat Gurinsky

Play now


On this episode of Network AF, Avi is joined by Senior Network Engineer Cat Gurinsky to share her journey through networking. Cat found a passion for automating deployments and troubleshooting and is the current chair for the NANOG Program Committee.

Today, Cat discusses her path into networking and her passion for automation as she shares one of her favorite things to do is write Arista API scripts and how she automated her process. You'll also hear insight into how to get started in networking, highlighting the importance of an intro class and mentorship. Networking isn't the only thing Cat is excellent at and passionate about; she tells us how she got started in martial arts and how it's connected to her networking life. Listen now to hear more from their conversation!


Hi, and welcome to this episode of Network AF. Today, I'm talking to Kat Gerinsky, an amazing networker, friend of mine from Nanogg Community.

We're gonna talk about her passion for Anume, Japanese, karate, networking, and we're gonna talk about automation, some of the things that she's been doing with it, advice for people looking to get into automation and a device for people looking to get into the field of networking.

Today, I'm here with my friend, Kat Garinsky, who's been in networking, and we're gonna talk about community technology and maybe, some non networking things as well. Hi, Kat. Could you give us a brief introduction to, you know, yourself, personal on work. Hi. So my name's Kat Karinsky. I've been working in network engineering since about two thousand and seven.

So good, like, fourteen years now.

I worked in computer repair before that as well. I was in college.

I am a little odd though because I'm a Japanese major for my bachelor's in masters, so it doesn't quite, you know, compute when you're putting that together.

But I was always really into computers, early on. My dad was actually the computer lab teacher when I was growing up. So it kind of ended up being a natural fit for my actual job, and it's hard to get a job as a translator, which is my original dream as a child.

It also doesn't pay nearly as well. So if I did get a job as a translator, I would not be on the same pace. Yeah.

And, yeah, and then I own a Cruddy dojo in Austin, Texas. And I have two small children, and we live here out in Austin.

Cool. I did not know, that you had wanted to go down the translation route. There's a I have a friend and amazing woman I went to high school with. You retreat to is Israeli, but, actually wound up moving to Japan and then wound up doing Triway translation.

You know Hebrew Japanese English and now is expert at training dog trainers.

Because I think, yeah, there's only so many there's only so many, you know, manuals that one can translate in things, you know, in the markets.

So so you mentioned, computer repair and, hopefully not just, like, supporting people who's hard disks were full and they couldn't, you know, web browser's because of it.

But how did that how did you get into networking? Yeah. So computer was, you know, people like, I had broken hard drives or, you know, broken CPUs that were starting to smoke or whatever. I, actually in a computer repair school at a tech school during high school.

Oh. My one son's story. Yeah. So I knew that I was gonna go to college.

I was one of those honors kids, you know. Like, I was definitely on the college path. Right? But I also knew that I didn't go out to college and work at McDonald's.

I just that was never, like, me. I was like, I wanna have a better job. So I convinced my parents to send me to school week, which is where most like the high school dropouts went. Right?

But I was like, no. No. No. I need to go to tech school so I can get a good college job while I'm doing college, you know.

So, like, I was like, I can put all my AP classes in the morning and I go pick school in the evening. I had all planned out, and they they're like, okay. You know, you made your case then let me go. I did my junior year of high school.

So then I I worked at the IT department at my university for the entire time I was studying for undergrad and graduate.

But what happened was when I was working at the IT department, as an undergrad student, one of my jobs was a technology assistant as they called it, which was that we helped get students online during moving, like, like, freshman and stuff, and then just help them with connectivity problems throughout the year. That was our assistant role of the job. And, when I graduated, we only had a single network engineer that worked at the university, and he was looking to hire, a full time network engineer work under him. And now I didn't have any other engineering experience at this point.

I just had lots of computer repair, right, because I've done all this stuff with the IT department. You know, he'd see me help people get people online. You know, we we deployed a Cisco clean access. I don't even know if that product's still around.

I don't know what it is. It's an axe solution. So, it was something that you would log into, and then it would verify in your computer that you had, like, antivirus software that your firewall was turned on, like, kind of stuff. Making sure you met minimum requirements that you weren't your risk to the network to infect the entire human face.

Yeah. And I helped deploy that. A as a student employee. So, like, it was, like, the one thing working my favor.

But he knew, like, you know, from the years of my time at the IT department that I could learn, And so he was like, here, you can apply for the job. I'm gonna be going interviewing you anyway. He's like, go study some stuff so I can ask you some relevant questions, you know, from the Texas book. He's like, but you know, I know you can do it.

So it's it's pretty much a shoe in. Right? And I'm like, okay. And so I got hired, as a full time, like, you know, on staff at the university, but it was basically an apprenticeship Because I I'd never, like, written any Cisco code at all at that point.

Like, I didn't have a my my clue what to do with a switch or an an access point I only needed to deploy that one server, you know. In fact, and Mac were fuzzy concepts.

Yeah. L two l three. What does this mean?

How do how do I terminate coax and ethernet and fiber? And, yeah, I actually learned all of that stuff at that job. So it was really cool. Like, you know, because it was, like, an apprenticeship, and those don't exist too often these days, you know, but I basically got an ever can share your apprenticeship with him.

And we worked together for, like, about a year, year and and then he left to go somewhere else and it was just me. Scariest time of my life. I stayed for another year. He did the boss I had and left.

So Yeah. The boss I had didn't really believe in me was the problem because I'd only been there as in that role for about a year before my boss left. So So it was more just is more just lack of experience than, you know, than you not being, like, you know, his expectations of of Well, yeah. And he he felt my lack of experience, meant that I couldn't possibly know what I was talking about either.

And so he never took my advice on anything. Even though, like, Simon, my my boss, that had trained me, you know, in that apprenticeship type role, Simon had trained me very well. But he just could not believe that I actually knew anything really. And that was kind of the impression I always felt, working with that that other boss after Simon Buffney.

And so that got really frustrating pretty fast. And, Simon, as it turns out, was friends of Christmas later. They had gone to school together. So this is where it all ties in.


And Chris was like, Hey, Kent. I'm just gonna take you. And you're gonna come over here and work for switching data with me. And I'm like, okay.

Please save me. Thank you. So I met I met Chris through Summit because they were like really close friends, you know, And that was how I got out of that situation and how I ended up at switching data, which went through the merger about a year to half into that Yeah. To Equinix.

So and that's how I found Nana too because Chris said, come on. Let's go to this conference. You know? Awesome.

Yeah. I mean, my experience has been that most of the people, in networking, especially when I was starting, were sort of self taught And so understood that, understood that everyone was learning all the time. Mhmm. So the only way to become expert was to go break it and figure out Yes.

How to unbreak it. Oh, oh, yes.

So, so true. I remember so many times that I'd be like, trying to troubleshoot something at the university when I worked with Simon. And I'd be like, Simon, how do I do this? He was not that I have a person that would just tell you how to do it.

So he'd be like, go look it up. You know, like, put but it's probably easy. Go look it up. Like, I remember that very vividly, my first year employment.

And so many times, I would get so mad at him, but, like, it made me so much better because I learned to look it up. I learned to try to fumble through it, you know, because at the end of the day, we worked at the universe, city. If I did break it a little worse in, like, three in the morning, no one's gonna notice they're all sleeping, you know, and he told me I'm breaking before they all woke up, you know.

I, the the wizard, Steve Robinson, who worked with my father doing medical research who, you know, sort of was, like, don't wanna do that basic stuff. You wanna use Unix and see and this is like in nineteen eighty. Mhmm. But when I I would do things, like, I would go to him and say, so I wanna I wanna, you know, do a science project. Let me, you know, see if I can recognize images with the computer. He's like, that's actually a much harder problem than you think. I'll tell you what.

And he picks up a piece of paper and draws and punches it with a pencil and says, go look at the world one pixel at a time and tell me what you see. Think about it. Come back and have questions for me. You know, but yeah. Or when I asked him, like, how should I administer this UNX machine? He's like, well, run fine slash and figure out what every file does and ask me if you have questions.

So but, you know, you do have to have time and privilege to be able to get you and learn that way. Right? You have to you know, I I remember yeah. I mean, I I had similar in in college.

I had done a bunch of computer stuff, but I met Actually, the guy that did image processing and the guy did networking. I was like, oh, networking. Yeah. Maybe, yeah, sure, whatever.

And then, but, you know, he was he, like me, would go to the computer shows, buy parts, build computers. He was doing it because we were at a state school that had no budget for so he was you know, like scrounging lab, and I was building computers for, like, law firms and stuff. Mhmm. Yeah.

And then so, you know, yeah. So then we started I got interested we had this empty thing called a t one, and then I had to figure it out. Oh my god. You know, my high school got a t one.

It was the biggest deal. Oh my god. T one to the high Oh, that would've been fun. Yeah.

That was, like, probably what I was I think I was in, like, tenth grade because I remember taking my c plus plus class and, like, we'd just gone through that year and You're gonna laugh at this one. Plus plus in tenth grade. Wow. Mhmm.

Yeah. I took a C plus plus class in the period, but I remember, like, as it was I, you know, it was a pretty easy thing for me to take that class. When we had free time after we did our program exercise, all of us would go log in and play neopets. That's gonna age me because nobody plays it anymore.

Well, but but we was at home. We had dial up. All of us had this horrible connection. So you could never, like, get the really fast things you had to click for loading quickly.

We could do it at the school at the t one. Yeah. I'm I'm having a vision of dialogue pixelated. Whatever.

What is it neopets? Neopets? It's like, like, a Tom I got here or something. It's like, it's just like, you know, you have to talk about, like, I'm sorry.

We have to, like Sounds like we should go making an escape and, you know, and and They had lots of mini games that you play to help, you know, get stuff to to make your neopet better. But, yeah, it was a very silly game, and I don't think any I think it still exists, but I don't think anybody plays at mainstream anymore, like Okay.

Okay. I predict I predict within a year we'll see it, you know, NFT, you know, on the crypto space. So so you know, you you talked a little bit about how you got in, and it sounds like, you know, you're fortunate, but also, applied yourself and prepared you know, even back from high school, and I will say, by the way, that I also when I was in in high school, there was, you know, people that owed tech school, but I I will, you know, it's worth noting how much more money plumbers and electricians make and carpenters than many than many I have a lot of reasons not go.

Yeah. Yeah.

And then frustration, if you find people that are maybe expect people to look like, you know, look like you know, to already have experience. I posted something on LinkedIn. I think we're looking for fifty six years of firewall experience, you know.

Yeah. Actually, Rachel Bicknell had found it, and I Oh. Because I commented and then and then reposted it. It was, like, so we're like, no.

No. No. They mean, you know, base fifty. You know?

She finds some really funny job postings I have to say. Yes.

Yes. Absolutely. So so What is the biggest outage you have caused without automation?

That's funny given the recent, basics. Yes. So, I was working we were think we were still switching data yet. I don't think we had merged to become Equinix.

I believe we're this is shortly before the merger. I was tasked with, collecting a bunch of information from show commands. I can't remember what exactly I was inventory, but I was basically supposed to inventory stuff. And all I had to do is do show commands all day long.

For this project for, like, a week. Mhmm. And, we had an extreme, I think it was an extreme networks, chassis switch in the Palo Alto Pax data center.

And I was just going side by side. And because we were all remote, you know, one of my few jobs where we remote before COVID.

I, you know, could do some of it during the day, some of it during night. It didn't really matter what time of day I did it. They didn't care as long as I got it done in the time allotted to me. Right.

You know? So I think it was probably, like, I don't know, eight or nine. It was sometime in the evening, like, after normal hours, and I'm just like, okay. Let's continue working on this because I got some free time.

And do my show commands and the switch rebooted.

Oh, no. It gets better though. Right? So if it just rebooted, that wouldn't be so bad.

It came back. Well, we had no consult to it. So I couldn't actually see that it came back, but, evidently, it came back without a config. So, like, I am, like, waiting for this thing.

Like, Yes. Pretty much. But I didn't write it. I literally had only done show commands.

It was some bug that it hit. Right? Anyway, but I'm, like, sitting there, like, I can't ping it. Like, it's not coming back.

I'm, like, freaking the f out. You know, like, what what's going on? And I'm only, like, been at this job for, like, maybe a year or less at this point, because I was only a year and a half before the merger. I remember it was before the merger.

And so we have to, like, call the poor one look. They only have one engineer that actually lived in the powwow area. So I'm like, Helen, Helen, I need you to go down to the plexig Hope.

And yeah. So he had, like, wake he was awake. He wasn't that late, but still he was passed aft after hours. It was dark.

He had to get up and go drive down. And go fix it for me by hand. Now thankfully, we didn't have any of these locked out security issues like that. This is just an internet exchange switch, you know, for the packs, but still, Like, let's switch seat down for a couple hours until, you know, he's able to bring it back.

No. I didn't lose my job or anything, and it was fine.

Cause it other than by tickling bugs somehow that Yeah. I just triggered a bug. Like, I didn't I didn't, like, duh, I didn't do a change. I didn't make any changes.

I literally was only doing a show command, but man, I was, like, freaking out for the good couple hours until that thing came back and he manually reconfigured it because I I've had a few people who've worked for me before, but they're like, oh my god. Oh my god. I just it's like, awesome. Now you get to fix it.

They're like, no. No. I broke it. I'm like, yeah. That means you get to fix it.

Live, like, in Northwest Indiana at the time, so there's definitely no way I could help. Like, I was way too remote, you know.

Oh, that was interesting. Right. So that was really remote. Yeah. For for the time. I mean, indeed America Verio was remote. I mean, some some companies were pretty remote at the time, but, that's one of the questions that really interesting is how do we help people lab and understand how things break and how to fix them without actually breaking the internet and and people's applications.

Yeah. That's something that I was talking to the Uber driver about yesterday because he's like, what do you do? And I said, oh, we make the internet go. Oh, I have a CC NA and it's expired.

I'm like, don't worry about that. Don't don't don't worry about that. Like, don't worry that it's expired. Like, go play with, you know, virtual virtual stuff and you know, show and I told you know what?

I I told him because we were in I was in Virginia yesterday.

Since I I'm getting back, like, three years ago, I have to remember what's City I was in, you know, this morning.

And I was like, you're dry and he was driving me to Equinix DC three. And I'm like, just look, all these buildings, they all have operations And if you have bright shiny eyes, you know, they can, you know, there's networking, there's cabling, there's, you know, just support and operations, and, yeah, a lot of it is centralized. But, like, if you're really interested, you know, because you wanted to not be in Uber driver rooms, like, just apply you know, apply to all these, you know, apply. And, you know, I give my email address and, you know, good luck.

You know, hopefully that will happen. So So I mentioned automation because, you know, automation can be a good way to, help humans be lazy. I I always try to be like, wait, why am I doing this over why am I doing show commands over and over and over? Right.

Why am I doing, like, the same type of thing over and over when I can automate it? Right.

But if you, if you, automation is also a way to break things really fast. Yes. If you don't think about it well. And and and further, you know, we see a lot of of our customers are like, oh, I'm so behind. And, you know, because the vendors are like, oh, it's all self driving networks, includes loop automation, and magic models, and intent.

And, you know, most of our customers are, like, if they have life cycle, like, I turn it on, turn it off, inventory. The interface descriptions are set, like, basic health. So, you know, then then they're doing pretty well. And so this has been, you know, focus of, of yours, you know, Python and and and is it more internet internet side or data center or, you know, what kind of Yeah. So focusing on. Yeah. So for me and my current role, It's definitely data center side, because I'm just doing really large scale deployments.

You know, I come through the switches instead of, like, you know, a couple.

And so it's It's both the automating of the deployment.

So a lot of, you know, Python scripting and power on auto provisioning and answerable usage and all that stuff, just to make life easier, including, like, you know, automate even assigning IP addresses, etcetera, too. Like, the more again, the more you, like you said, the more automate that gonna do by hand, the better, because I can, in theory, reduce human error risk.

And then then after it, you know, things come up too, I give matter of, like, automating my troubleshooting as well. Like, some of my favorite things to do is to write, like, like, arista API scripts. I love their API. It's really easy to work with.

And if you know Python, It's just all interactive Python with our API. And I guess at this point, we're like, you know, okay. Every time a link has errors or there's some sort of issue with the link, like, I'm gonna do the same ten show command. Right?

Mhmm. And so, I just was like, why don't I just automate that? You know, and I can do that all through the API and just grab all of that data that I need every time. If I'm gonna be doing the same commands, we'll have the API grab it really quickly.

You know, that's I physically cannot type those commands as fast as the API can can execute in. Right? You know? And then they're always gonna be the same.

All I have to tell it is host name and port number, you know, making my life way easier. So Is there any rate limiting concern? Like, what if everything started showing, you know, interface? Does does the risk to do its own rate limiting, or did you have to implement that yourself?

I don't think there's one that I've been able to hit yet. There may be a rate limit, but I've not been able to hit it yet. Like, one for example, like, one script that I wrote, which is kind of in between deployment and troubleshooting is, I got really tired and we would deploy, like, a ton of servers on a whole rack and then having to go back and put port names in there, because I didn't like having, like, ports that didn't tell me what server they went to. Like, I really wanna know what does this connect to.

Right? So later out of somebody's, like, saying, hey, my server's having issues. I, you know, yes, Trace strat will tell me where your port is and, you know, show Mac address will show me where it is, but it's even better if I can label the port. Right?

And so I wrote this script. Like, the first thing I wrote actually when I took the Arissa API class they taught. It's back in two thousand seventeen, so four years ago. I took that class and had that little light bulb moment.

It was like things I could do with the API. And that was the first light bulb that went off. It was I could automate putting port descriptions on switch ports. Like, between LLDP and ARP and IBP six neighbors, like, I could between one of those functions, I should be able to figure out what connects here.

Right? Mhmm. And instead of me having to do those three different commands to figure out what connects here, could just automate that. And so that switch that script that runs really fast.

Like, I can run it against all, you know, all the ports. And it'll just, like, and so that's I can't think of anything. I would run that would be fast. That would be that would be more requiring a rate limiting, really, than something like that where because bam bam bam bam bam.

Mhmm. But when I run it, yeah, it just Cool. It goes quick. It does the IP lookups and everything if it's ARDP and everything really fast and the switch replies with everything.

I'll do the API and then update studies for me. So what's the in this case, the source of truth is the so you're rewriting it. It's not a database. It's not like a net box.

No. It's not a database. I'm doing, like, an asshole, basically. I'm sorry? Yeah.

Yeah. I'm doing an as built, basically. Like, it's like, this is the reality. Yeah. Okay. So I'm not I'm yeah. We don't keep a database for what port that's wired to what server.

So, yeah, like, I I try via a couple different methods. I usually preference our our LDP first because that's very, very reliable for the port. Right? If it's a port channel, then I'll be assuming the preference ARP, if I can.

And, and then if it's a v six neighbor, then I gotta do v six neighbors instead because you're not gonna have ARP obviously on there instead. So I kinda go in that pecking order, though. LDP, check ARP, check v six neighbors if it's not a p four address. Yeah.

So it's literally it's it's definitely there. And hopefully no one has jacked a switch into the other ends. There's many arps, many things that you see on the on the, you know, either ARping or or Mac learned, you know, speak to the other. They should just be accessed, I guess, if they're all Right.

Well, yeah. Exactly. So since it is a data center, it's a very standard build as far as how the ports are laid out and what's going on them. And I've I've never ran into that in the data center world.

The enterprise world I used to do that. Yes. I wrote it all the time. People plug in hubs and other switches and things into stuff they shouldn't, but the data center side, that's really not a non issue for us plus our our are gonna be on QSFPs and our servers are gonna be on SFP or on copper.

So, again, it's you can't accidentally plug into the wrong thing because it's a completely different, you know, sounds like a that sounds like a design feature for sure. Yeah.

To have physical incompatibility with with things you don't want to, if things you don't wanna plug in. And, yeah, I mean, we see that, you know, we have some features we built that assumed that people configure things with reg x. Like, you could reg x match, you know, interfaces to find, you know, server dash, whatever appearing, whatever the people do that. Then you find, definitely some networks that are have organically evolved where, you know, in the worst case, if there's an interface subscription, it's wrong because people just gave up.

Yeah. And so then this question of what's the source of truth, you know, in in a non well formed data, you know, center will can be interesting. You know, people try to have what but sometimes the iPad, the CMDB, the configs, and they're like, oh, but it's some of it is true in each of those. And it's like, no.

It's starting to One thing I have to always be careful of is, like, for that scripts instead of doing, you know, what's actually connected, right, is that we don't wanna run it too early. So we wanna audit and make sure that people plug things in correctly, like, uplinks, for example, right? So, like, if they had we usually, like, four uploads on our switches. Right?

So, like, maybe the first two came up, but the second two didn't because they accidentally flipped those when they were cabling them for us. Right? Maybe core three and core four are flipped on the switch. Right?

Well, if I run the strip though and it looks at LDP, it's gonna put that wrong label on that wrong port. Right? And then suddenly, I think it's supposed to be like that. Know, so it's like one of those.

Make sure, like, all your links are actually up before you run the script. Make sure it's correct. You know, like, sir, you gotta get a little bit. The service is less of an issue because they're access on the same VLAN, you know.

It's all. Yeah. That is an interesting yes. That's interesting. If the source of truth is the network itself, Right.

And you start labeling it too early while it's still in flux, then it could confuse people who, right, who are trying to operate. Right? Who didn't And so then you have to rerun the strip later, but now you put that little in between ten. Right?

You've got that wrong label on there. So that's one thing we try to do with bring ups at least is we wait till we verify all the PGP up all that stuff goes where we where we ask for the patch plan for the switch to switch stuff. Right. Now the server is, you know, there's it's really hard for us to tell if it's actually done correctly.

Really the only thing we we have a a different guy who wrote a script to look at the database, the LVP. And, the only thing we really audit on our redundant switches where we have, you know, two in m like Paris, we will compare and make sure that the, l d p is the same on the same number ports. Right? So if we see, like, port three has server a over here, it's on port four.

Okay. Then, obviously, somebody cable's going wrong. You know, it's like that's, like, something else we've gotta to check for because then the LA is not gonna come up because they're on the wrong ports. Yeah.

I mean, it it's interesting. Some of the the table state can be really useful. People are I I I I I know a couple companies that have made streaming telemetry of the blurb everything work, but Mhmm. It's really hard versus just, you know, do the show things match up.

Yeah. Yeah. Exactly.

No. I'll have to I'll have to see if I can torture the ARRISTA. So far, they've done everything I've wanted rate limiting wise. Exactly. Yeah. I've never had a bar from this. I had a wrong password or something.

Or I just did the wrong command. Yeah. No. I I got so annoyed with, I have a nine thousand one that I put in the CEO cabinet because I still have a three p x l that's got slash twenty four is filtered there.

And I was like, wait, I have to Google this just as much as if I were, you know, sticking at MX, and I think I should just go with it. And I use the wrist stuff for someone is help them set up a lab of, like, data center. Like, you're, like, you're talking about, and it was, like, Wait. This is just iOS with commit confirm and bash.

This is, like, much better. Oh, I love it. Yeah. Like, I can't go back now.

Like, I'm so spoiled. Like, I just I can't can't go back to something else. I I like it too much. It's just it feels like home when I'm on that box, you know.

I mean, there's a lot of def definitely a lot of juniper bigots, but I'm just personally, I'm just slower when I have to and and I'm always doing displace set and they make fun of me. And it, like, oh, but I'll be, you like decoding. Why don't you like braces? I'm, like, oh, but I'll be, like, see coding.

Why don't you like braces? I'm, like, That's for programming. This is for router configs. Like, they're Right.

And I do like programming. You know, I like my well, Python doesn't have any of the bracket stuff either, but, you know, but, yeah, it's it's I can I'm I've not done that kind of programming in so long that it's just easier for me to stay on the other side. Yes. So so it sounds like, you know, I would say You're doing stuff that's, especially, you know, with the show and augmenting and some diagnostics.

It's probably more than even, the average, you know, that we see.

Where would you advise people to, like, what's been helpful for you to try to, to try to learn, to use virtual ARristas, a particular place on, you know, their libraries you love or their, you know, or, you know, anything in particular that you can remember that you'd point people towards if they're looking to. Start doing the automation dance? Yes. So I definitely recommend, like, taking some intro classes for sure, just to kinda, like, get your feet wet because I'm I'm a good self starter, but I always find that, like, that initial, like, you know, dipping your foot in the pool type thing helps a lot with the intro class.

And then once you have the feet, and then it's easier to to simmers versus the body. Right? Mhmm. So for me, like, I took the actual Arista EPI class that they were offering.

I think they still offer it now or at least they did pre pandemic You know, if you're looking to do, like, something with, like, ansible and take an ansible introduction class, you know, like, do do those intro classes to at least get the framework before you just dive into the deep end, open up the books, and go looking for this exact type type of thing you're doing. And, like, for me, I did help that, you know, back in high school where I had to that c plus plus class and then during, college. I did a lot of, like, HTML and CSS stuff. So I had a kind of framework.

I did a lot of self learning for PHP at the beginning of my network and training career. Well, all my scripts used to be PHP. I actually didn't switch to Python until I took that, I erased the class. And they're like, well, you're gonna have to learn Python for this.

And I'm like, Okay. I guess I'll start reading those. So that I did self teach. But then, like, you know, the examples they could taught us in the class to interact with their API helped, like, not baseline.

I got it. Okay. Here's the basics of Python. Here's how they're actually doing it, and then I was able to to learn more from there, seeing it all put together.

But, yeah, definitely intro classes, I think were the biggest help for me. And then as far as, like, labs, I've been really blessed that we have physical labs here at my current job, but, obviously, a Q and S3 with, like, you know, virtual EOS image or something would be the way to go if you don't have that, which the average person probably does not have a physical lab. I happen to work for a very large company. So there's the Yes.

And or or, you know, it's it used to be people with eBay, you know, stuff and try to make their own That does work too. Yep. Yes. But virtual is pretty good.

Yeah. I was surprised when I looked at the arista side how much well, if you haven't ever use the CLI, you probably should do that before you try to do automation.

You know, there's a lot of Oh, intent and models, but, you know, when you actually look at how you automate the major vendors, you sort of do have to understand how how it thinks of itself and CLI entity and stuff. So And you have to learn how JSON works too because you need to understand how to read the JSON dictionary. So much of what you get back from the show commands on API queries are gonna be, like, in JSON format. So that's something you need to quickly quickly learn and adapt to, to understanding, you know, how to find your nestle's variable inside that show command, because it's give you the text.

It's gonna give you a JSON version of the text. One cool thing on the arista is, is that, you know, you can do your show command on the actual arista, and then just type it JSON. And if it's supported, it'll give you what that JSON format looks like on the switch. So you can just debug.

Yeah. So you could. Yeah. So before you go to write a new script. Right? You know, I, okay, I wanna use something where I'm gonna do, like, you know, show version.

Right? But, like, I don't know how the JSON's gonna look, and I don't wanna start from the script. Right? So I go to switch.

I type show version pipe JSON, and then I see right there on the switch what the output's gonna look like. So now I can go back and write my programming based on what I know to expect. Does it show you a queued, you know, like, like, pretty printed, or does it show you, like, all, like, blah? On the switch, it'll be pretty print by default.

Okay. Yeah. So it's it's nicely. I'm trying to figure stuff out. And I I still use Pearl because someone's, you know, even though it gives me a little bit of a headache to navigate the hashes that Jason parses into.

I I need to get with the Python at some point. I just think the white space shouldn't be syntax if we should have braces for coding. We should have braces. Oh, I know.

That part does does still bother me a lot. I agree. I'm very used to, like, you know, I did at PHP, very stupid brackets to tell me which blocks belong where? But we just did over time.

Yeah. I can't fight the industry. The industry has decided Python. Exactly. And that's where I caved.

So in your background, you know, you were talking about sort of, you know, dipping your toe in and and before, you know, your whole body and I remember my experience, you know, getting into martial arts where not only I, but everyone else thought I was just gonna turn I was turning red. I was three hundred and twenty pounds. People thought I was gonna die. Oh, wow.

But But, I just, you know, for me personally, I suck at it so badly that my it takes all my mental focus And then I don't notice some exercising. And then I'm learning, you know, and I enjoy things that I suck, but at. Yeah. But I'm curious.

So I guess you you mentioned you have Japanese, you know, Japanese studies, and I guess that that's some relation, but I I'm really curious Is there a connection, you know, between martial arts and networking, anything that helps you from one to the other, or is it more of a disconnect You know, how do how do you find it and think about it? You know, as I know, a few people in the community, they're into martial arts. And and you you're you're you know, sort of actually have a school. Right?


Yeah. Crazy. People like me opening schools during pandemics.


Well, I got into martial arts before anything else, really.

So the Japanese and the martial arts definitely relate heavily for me because I was really big into anime. It was one of those Dorkio Taco kids growing up, like skinny and lanky and getting picked on. I just live watching my anime, you know, And so, one of my friends actually found this, karate school that had a sword class in the Idaho sword class. Oh, wow. We can go take sword and learn to to be like tension.

That was what she told me, sold me on that. I'm like, oh, sure. Let's go learn to be like tension from the anime, you know? Wow. You know, then eventually after, like, summer of doing so, they convinced us to stay for the karate class. And so that's that was kind of the beginning of everything. That was two thousand, so, like, twenty one years ago.

But, yeah, so that that's how those two relate for me is my Japanese interest came from the anime. My martial arts interest also came from the anime.

I mean, there's definitely relations. Right? Because I mean, martial arts to me is kind of like another language. It's a language of the body, right, versus, like, you know, networking is a language of the switches, and then the programming is, you know, the language of the code to interact with everything. So I think there's some some similarities, but it is also my break from the day job as well.

So he's when I sit there and program all day, I don't move from this chair very much. I'm very dormant on this chair for, you know, hours at a time. And that's the opposite of when I'm in the dojo moving around constantly, like, no no downtime. Like, I have a chair and a desk at the dojo, but that's really only for when it's not class time.

You know, if I get there early, I need to update the website or something. Like, it's about the only time I'll use that. The rest of the time, like, I never sit in that chair at the dojo. I am up on my feet whether I'm talking to a parent, talking to a student, So, yeah, I I don't know.

Terminatives. It's interesting because in networking, I, you know, people are trying to get a networking. I say, the most specific route always wins. You could remember this because if you couldn't get to your own if the default route if the biggest route won, you couldn't talk to yourself.

Because you're, you know, in in martial arts, it's like remember to breathe. It's like, I know how to breathe. Like, Discover, maybe you don't know how to breathe. Yeah.

Because you're I try to remember this so hard.

I did I did that. I'd almost feel like Avi, you're gonna pass out. I was like, oh, shit. Maybe I don't know how to breathe. I should think about that. You know?

And We definitely go back to basics a lot too. Like, people don't realize, but there's so much focus on basics and martial arts, you know, just like your your basic stance or your hand is, your positioning, you know, how you're leaning into something. If you're back straight, if you're squatting down low, lots of, like, fine details that that matter martial arts.

That is one of the things that I find fascinating myself, which is that I can see how people that can do things much better than me are doing it not quite right. They don't I'm not I don't give them my opinion, but They're right. Know, but I'm still sucking so badly. I'm so far behind.

Like, I remember the first time I could make my body do each of the each of those, you know, sort of basic things. And then when I was starting, it was the ukemi role, which I really sucked at, but, and it'll still do a little bit, but Yeah. And, you know, the internet and networking like the human body, you know, there's so much that can go wrong. So luckily, I haven't injured myself at martial arts and, you know, hope everything is good.

Your school from that perspective. Yep. Same. Cool. So we talked about automation, which is one of the things that I hear a lot as people saying, you know, oh, I'm behind and this is this is the hotness and everything else. And, you know, then I say, well, you should be thinking about it if you're not, but, you know, with the people are not living in this completely automated world with no humans. So that's all APIs.

What else is you as you look at networking? Is there anything that you're really looking forward to immersing yourself in between, you know, self exploration and classes and and things, any anything you're following that that that you'd like to be working with?

My kind of big self driven thing when we separate from automation is IPV six.

Big, especially if I'm looking at large scale data centers, like, there's just so much IP address wasting in the v four world. Like, your points points are my favorite example of this. Right? You know, like, my typical install for an edge switch, top of rack switch is gonna be four uplinks.

Right? So four uplinks means that I have eight IPV four addresses. That's better than it used to be when it comes to slash thirty. Right?

Yeah. Yeah. We are doing yeah. We are doing thirty ones, not thirties ago. So I could be wasting twice a second.

At least I'm doing that. But even that, still eight addresses for every single switch I wanna deploy. Minimum plus the loopback, but the loopback side don't care as much. It's more of that.

That's that eight addresses per device And when you're talking about deploying, you know, a hundred devices, right? Like, there's eight hundred IPs now. I just wasted.

Even if it's internal only, there's only much internal, only space I can use as well. Right? You know, it doesn't even have to be public space. Right?

And the v six problem just eliminates that as being an issue of worrying about running out of space, like, you know, and so kind of one of those big hot topics that I'm trying to work on is at least start with the networking side. Right? Because not all the servers support v six. Right?

That's the other problem. Right. And then let's go back and obviously a server issue because that's the way bigger one. Right?

Obviously, I've used way way way more IPs on the server side. It's not just servers. It's applications. And Yeah.

Server's applications. It's different. Yeah. Some servers might have like ten IPs on them. Yeah.

So, but, yeah, my favorite thing that I can do on the wrist is right. And I think of this, which is do this too, is I can do v six routing over or v four routing over v six interfaces.

Mhmm. So even though the whole path is not v six, I can still send all my v four routes over v six only point to points. Right? And so, Tada, now I no longer use eight those eight IPs of v four at all.

Right? Right. No. That's So that's gonna yeah. One of my good projects. Does the when you think about making it more approachable to crafty people who are like, like, oh, but I can't remember v six addresses and all that.

Is is Is it, like, having scripts for debugging or tools so that so you don't have to, you know, be, like, typing these when you're debugging, you can say that switch, that interface, like, do some debug. Exactly. Yeah. Is it ring so that, you know, before any v six are related, or there's some consistent like that switch, that interface will be this.

Like, what are the things that help crafty people? Like, like me, maybe, you know, wants want to be in that world?

Yeah. So, like, you know, it's slightly wasteful, like, the thirty versus thirty one argument. But, like, you know, I do one twenty sixes instead of one twenty sevens, for example, in v six. Right? If you do that and you have four uplinks, then you'll use the same four addresses for your first core, second core, third core, and fourth with all, and then the same number. Right?

So that's one of those little tricks that we do. For example, again, in that four environment, because then it follows a better pattern that's more human readable.

And then, yeah, we do a lot of automating for our our point to points and our basic everything, basically, I've I've written scripts that will automate making the nice pretty DNS because I don't need to memorize the IP if I have DNS for everything, including my my v six point of points. So then if I do a trace route, it may sense because I can see the host name with the interface, you know, attached to it Mhmm. On that trace route. So but, yeah, DNS, for sure.

Well, if it isn't a conflict of interest, I would be interested in, and and if work allows it, maybe an analog presentation or maybe a panel on you know, with a few other companies on, like, cool tricks to, you know, digit and at least for network interconnects, you know, internally.

Move away from, v four, which, yeah, if you're only doing twenty switches, maybe your beam counters don't care, but if you're doing you know, make money. Hundreds hundreds. Yeah. And then it's money, you know, it is money. Yeah. Not to mention of being nice to people that still have the mainframes that need to connect and, you know, need to work that way.

Cool. So, you know, definitely a lot of people passionate about that. And Hopefully, I'll I'll be able to attend the panel and ask a couple questions.

Mhmm. So you got you got, lucky on top of some applied work getting in, to, you know, full time IT and then networking. And moving over to, the multi tenant networking, you know, the switching data side. Yep. Any any advice or or things that you think about as a community that we can do to help attract people, especially people that are not, you know, a diverse set of people, you know, into networking.

Make it more approachable and and offer opportunity?

I mean, for me, like, mentorship really was the key. Right? That was the thing that opened all the doors. And so I think the more we can do with mentorship the better for sure. Like Nanock, you know, which I'm I'm chair of the program committee over there. The board did start a mentorship committee, and they've been working on on ways to get that going. I'm not sure when it officially goes full deployment, but definitely, like is is is a priority.

They don't give us a good way to learn some stuff, but this is a specific, like, committee just for mentorship where, they work on, you know, it's formulating a whole plan for how we can mentor, you know, up and coming engineers. Right? And stuff like that, I think, is super important. I have a huge passion for mentorship because it's really how I got started.

Right? So I definitely think if you're starting to get interested in the the networking engineering world, to find someone that that can mentor you were to ask around and see if you can help find someone to pair up with that it can give you advice. Can, you know, introduce you to other folks in the network world. Right?

You know, Chris Miller, like, knew everybody then. I'll go ready when I started going, you know, ten years ago. Yeah. And I wouldn't have met half the other people like yourself and others, you know, I better have somebody like him introducing me to all those other people.

Right? Yeah. So, but yeah, I get having someone to to to bounce ideas off with to learn from, to help mentor you, career wise, actually, you know, in network interior itself or, the technical details as well. All of that's super useful, I think.

And that's how I think we became approachable, though. So likewise, if you are considering being a mentor, right, now is the time. Like, don't wait until somebody finds you. Right?

You know, people like myself and you, we need to go out there and look for the next, you know, generation of engineer.

Yeah. I mean, we're thinking a lot about that in terms of college. You know, how how can we help because we have a lot of data about the internet. We can work with grad. School, you know, we can talk with people even that are not technologists because, you know, with peering, it's not all about technology. It's about politics and economics too.

And, you know, try to find the bright shiny eyed folks.

We haven't yet stolen someone, you know, convince them not to go back to college. You know, we don't Right. We didn't about that, but you know, I can't say if we found someone, you know, who's right, we would at least, you know, give them the option.

But, you know, that's that's one of the areas where, you know, the education and and labs and things, you know, as you mentioned, it's possible to do online. Know, we're thinking about is there a way that we can help with that? Or, you know, we have some traffic generators help, you know, so that if you stand up a network, you know, we can help, you know, put traffic on it to show that it would look like or, you know, on the internet side.

I need to look at that mentoring, thing that you just mentioned, well, we'll see each other in a few days physically and not logging.

Because, you know, even Andre mentioned Andre Tungku I I talked with mentioned that At times, even Nanogg was intimidating for him if you get trapped between, even two very nice people if they start, you know, fly spittle flying arguing about, you know, arista versus whatever, or, or, you know, this library versus that, or this architecture versus that, we have subodentic passionate of folks. But Nina also found, you know, she just, like, showed up. Someone showed her around, introduced her around. She found the community very accommodating, but you know, I think there are a lot of people for whom it still could be intimidating to, well, especially post COVID, to go to a thing and then make relationships. And then, you know, so I I look forward to learning more about how people can sign up and, and offer to help mentor.

I mean, have at least having a mentor going to my first conference already like that made it not a scary thing for me at all. But, obviously, many people go to cop those conferences and don't have anybody they know. And that's a totally different experience, like night and day different experience. So When you all look to hire, you know, if we were talking about or two people that are early in career, you know, what are you looking for?

Are you looking for a degree? Are you looking for a Github? Are you looking for they have labs? You know, what what what makes what makes, you and your peers interested, in in a candidate.

Yeah. So, I recently had an intern last summer, and then we eventually hired him on full time this year. And, in that case, like, you know, when I was looking for from as an internist first, right, you know, I was looking for some basic knowledge of the the concepts. Right?

Was looking for some programming experience, which this person had, because I knew we want to do more with the automation. But mostly and this is it goes back to, like, giving it back right from what happened with me is I'm looking for someone with an aptitude for learning. Someone who's not so said in their ways that isn't gonna pipe me in a return and is gonna be willing to hear what everyone has to say on the team and then go research and learn and improve themselves. And and we found that in that intern who's now a full timer with us.

You know, he was extremely eager to learn, had a good solid knowledge of programming, a good solid knowledge of networking, brought new ideas at the table, which is always a plus I I love someone who can think differently too, personally. If I'm gonna have a new member of the team, I don't want them to think just like me because then we're not gonna grow. I need someone who's gonna say, what about this? I I love the what about this.

Even if it's wrong, but what about this still gets me thinking, you know? And, so those are the kind of things that I look for. Right? Someone who is has a fresh set of eyes on things but has good basics and then can learn.

And that that's the ultimate most important part, though, they can learn. And they're willing to learn and they're willing to take feedback.

Right. And maybe, you know, understand that learning is the thing which is our responsibility. Right? The company can help train, but you need to be you need to be throwing yourself into it. Exactly.

It's interesting because you said sort of be open to listening to others But, you know, have new ideas, but it sort of sounds like what I used to do, which is like, okay, I hear you, but I think my way is better. Let me code them both up, and I will show you I have and and and I had a coworker before this current one, that was here for, like, one or two years. And and I had to do that sometimes just to show him because he wouldn't listen. He was really stubborn.

It was the opposite experience, you know. Like, he was just convinced that his way of doing things was the only way he should do it. And I would even, like, code things and be like, look, see this other way works like, you know, five times faster. You're six, ten seconds, mine takes two.

And it was really frustrating that even with, like, evidence, you know, like, here's the proof. Like, try it a different way. Like, that that person just wouldn't listen. It was so frustrating.

It's so hard to work with that person. And, like, a huge breath of fresh air, switching to somebody who is willing to, you know, to work with me and and and be open to new ideas. And sometimes I will get it wrong. Right?

But and but that's okay. And I know that. Right? You know? And but somebody who doesn't think they can do any wrong, that's a dangerous thing.

I don't wanna work or things they can do no wrong.

Goodbye. You get super villains. You get superheroes that come into super villains when you have Exactly. And, no, we definitely look for that to try to screen out what I call geek binary it is. There's one true way. There is no grayscale.

Like, z I, you know, or, you know, silicon valley made fun of spaces versus tab. I was so happy when I saw spaces versus tabs. Means jam TV show.

Really, don't we have pretty printers? Can't we just, like, teach, you know, each work in our own syntax and meet in the middle, you know, Yeah. I mean, like, so the, like, going back to your original thing about being approachable too and, like, you know, getting new people in the field, right, we also have to be willing to accept that the way we, you and I have been doing things for the last ten or more years. Right?

It may not be the right way anymore. Maybe that was the right when we started. Right? But we also have to have that open mind too.

It's not just our new hires. Yeah. Make the open mind, we need it too. And so we get locked in that curmudgeon, you know, way.

That's not gonna help anybody either. So it kinda goes both ways. Right? We both gotta be willing to to learn from the old and learn from the new.

And, you know, it's interesting for people that are interested in making a company starting a startup it's that intersection of aha. There's a problem that isn't being solved. Maybe this different approach that people to say is crazy.

You know, is the right way to go. We heard a really amazing story yesterday at Kentek, one of our, investors is Angels, in this new round. So it started as l l b g t q plus, but, and you know, Lorenzo Thune was talking to us, and he, like, happened to randomly run into George Takai and his husband at a show. And then another show And then, like, they talked and then he turned around and said, can I write a a a a Broadway, you know, show for you?

And he had never done that. And I was like, oh my god. That's crazy. And he's like, oh, but you have to be crazy too.

You know, you have to be crazy. I'm like, oh my god. That takes that that's that's like, but you know, sometimes you say, like, wait, why why is it done this way? And that's led to some SDN stuff that didn't make much sense, like centralized flow controllers.

And it's led to some stuff that, I mean, look at VMware, NSX and VLAN, and I was like, what's wrong with VLAN? I know how to python that or pearl it or whatever, but it turns out that that most, you know, people with brand servers didn't want talk to network people and open a ticket and whatever. And so someone would have said there's no problem with that. So, yeah, definitely can be interesting if you wanna go that path too.

So it sounds like you're having fun. I'm I'm I'm having fun too or I'd be jealous, but any any advice you'd give, you know, earlier cat, you know, from, from anime, for fighting to, late high school to college to early career? There's definitely one, for sure, piece of advice that I I I would definitely give early cat. And that was, like, the my biggest regret from college times was that I was such in a rush to graduate.

So I was, like I said, I was an honest kid rate I came in to to class, and I had some AP five scores. So I had a couple classes that I got credit for in college. And then then I took one summer where I I I took two summers. So with a one summer of my, core requirements and another summer of Japanese, full immersion.

And between those two summers, I was actually able to graduate in three years instead of four. And I would go back and say, don't be in such a rush to grow up, you know, like, take the full four years. And especially in the the biggest regret of all of these is that I never studied abroad in Japan, and I I should have. You know, because once you start working full time, like, going overseas for a whole year, unless you're gonna relocate for your job, probably not happening.

Yeah. We have two good friends that did, you know, and one wound up living there.

One, she was studying pro divers, you know, for for and the other, you know, as I mentioned, she wound up living there and doing, and I think they both, you know, enjoyed it. We we love learning then. Looking for. Yeah.

And I'm who knows? If I had done the study abroad, maybe I would have ended up staying working in something related to Japan instead of doing network engineering. So my whole life might be totally different, but Yeah. That whole, like, don't be in such a rush.

Like, oh, because you can do it. Don't do it. Like, take that time in college to really explore those things that are of interest. And and just, like, study abroad if you want to.

You know, I should have graduated in four years, not three, but and I I graduated honors. I mean, I I did well in those three years. And obviously, I'm doing well career wise today. I definitely regret that I never studied abroad.

And I only went over for, like, you know, short two week trips, and I never did that full semester or full year abroad. So that's the big thing I would tell myself is, you know, if you were thinking about it, do it. There will not be time to do it later, you know. Later.

Right. Interesting. Yeah. I I've had a few people in my family who did the poof your doctor program, you know, which used to be you know, like, you you do med school and undergrad and and have had some feedback about, you know, maybe I wish I had more time to, you know, do whatever then I actually wound up getting in got getting into, I I if I had a degree, I would have an undergraduate major in classics, which people don't believe.

But I don't have a degree. So it doesn't matter. I, you know Oh, I know some people don't have degrees, or they're like me. My degree, it is absolutely nothing for my career.

Like, my Japanese degree is not needed at do what I do for my day job. So so why does it matter that I have a bachelor's and a master's. It doesn't matter at all. I don't use that bachelor's or master's for my engineering at all.

Right. Yeah. I'm just having a degree doesn't mean you know how to learn in an unstructured way, you know, which can be really helpful. So, Cool.

Well, hopefully people will, listen and and, and maybe that'll be helpful to them. And, again, I really enjoyed this so much. Thank you for joining. I look for seeing you in person, either tomorrow, in Austin or a few days in Minneapolis, such as the life of traveling network nerds.

Oh, even even in the, in the hopefully, winning days of COVID.

Yeah. And, you know, I hope, everyone enjoys this conversation.

Yeah. Thanks so much for having me on. At least you should.

Got a guest?

Network AF is accepting guests for upcoming episodes. If you’d like to be on the podcast or refer a friend, reach out to

About Network AF

Network AF is a journey of super-nerd proportions into the world of networking, cloud, and the internet. Avi Freedman, self-described internet plumber and podcast namesake, hosts top network engineering experts from around the world for in-depth, honest, and freewheeling banter on all-things-network — how-tos, best practices, biggest mistakes, war stories, hot takes, rants, and more.
We use cookies to deliver our services.
By using our website, you agree to the use of cookies as described in our Privacy Policy.