Dial Up Motive
Presented by: Dan Hollinger, Chelian Pandian
Originally aired on April 1, 2021 @ 10:30 PM - 11:00 PM EDT
Human-interest segment asking Cloudflare employees what their first internet experience was and how it informed them joining Cloudflare. Dial-up modems, bulletin boards, punch-cards, Twitch, Twitter and more. Chelian Pandian, Solutions Engineer for Magic Transit, will be the guest today.
English
Interviews
Transcript (Beta)
Good morning, good afternoon, or good evening, everyone. Welcome to episode 17 of Dial Up Motive.
I'm your host, Dan Hollinger. And on this show, we explore some of the early Internet experiences and technology experiences of our Cloudflare employees.
With me today, I have Chelian Pandian, Solutions Engineer here at Cloudflare, currently helping support Magic Transit.
Chelian, would you mind giving a quick introduction?
Good morning, Dan. I'm Chelian Pandian. I'm a Solutions Engineer in Cloudflare.
So I have worked in Internetworking companies such as Cisco, Juniper, and Arista in the past 20 years.
So being a network engineer all my life, being a product manager as well.
Currently, I'm a Solutions Specialist for the Magic Transit solution from Cloudflare.
Awesome. And we're glad to have you.
And you actually started a few months ago or a few weeks ago now? Yes, a few weeks ago.
But this is my second journey in Cloudflare. I was in Cloudflare in 2018, 2019 as well.
So not new to Cloudflare. So your work is so nice that we hired you twice.
That's what I'm hearing. I love Cloudflare so much. I came back to work again.
Awesome. You didn't have enough orange shirts. I didn't have enough.
Yeah, my children do. They love Cloudflare orange shirts. Well, we're glad to have you back.
And it's great to see that you're working on such an interesting project like Magic Transit, and especially with your background.
And, you know, thank you for volunteering for the show.
And, you know, happy to kind of begin that trip through nostalgia and through your early experiences with both computers, technology and the Internet as a whole.
So we'd love to kind of dig into your first experience with the Internet and what that looked like for you.
Yes, Dan. When I came to U.S., I came for my master's in computer science in the University of Missouri, Kansas City, in the early 90s, 1991 fall semester.
At that time, my first time I encountered Internet was when my college IT team gave me an email address and said, this is the email you're going to use to talk to your fellow students and your professors and submit your projects and things like that.
That was my first touchpoint with the Internet.
And I, you know, I started using the email application, which is totally text based during those days.
None of those graphical user interfaces existed.
So we were using mail clients from DECWAX minicomputers and Sun Microsystems Unix email systems.
So that was my first journey and touchpoint in the Internet.
And was that pretty foreign to you for someone basically say, hey, here's an email?
Did you really kind of have to think through how am I going to use this?
What is this? Why is it, you know, better or worse? Like, did you have to go through that journey?
Yes. I was wondering what am I supposed to do with it and then how I can make use of it.
And then the first email I sent within my family was to my cousin.
He was working for DEC, Digital Equipment Corporation in Boston area.
So as soon as I came to University of Missouri, Kansas City, and they gave me an email, the only person I knew in the U.S.
was my cousin from my family's side.
So I just punched in his email address feverishly and said, hi, Senthil, here I am.
This is my email. Please send me email. Just so you know, you're getting my emails.
And then he sent me a very cryptic message saying, ACK, A-C -K.
He was being a network engineer himself. Yeah. He didn't want to invest too much in the response.
So he's like, if I prove to you your email worked, let's move on.
Exactly. And I mean, was that an early struggle back as the Internet was getting started, the email was becoming more ubiquitous, was just finding someone else that had an email address.
Exactly. Luckily, I had all my friends' email address.
So we called each other, all the people who came with me from my undergrad school in India.
So I called them, got their email addresses, and then punched an email to every one of them immediately saying that, hey, here's my email.
Let's all stay in touch.
So that was our connectivity mechanism back in the day for all the students from India.
So it was great. I mean, we were all able to connect immediately and stay in touch.
And also when you log in and inside the school computers, when we log in, we had our students all spread across all over the campus.
And they would log in from different parts of the campus and we could easily get into a chat program with my fellow students in master's and we could immediately work on projects and things like that.
So it was all text-based, but it still solved a lot of problems for us.
Yeah. And did you feel that was kind of your first taste of the future there and understanding like, oh my gosh, I can have this kind of communication with people around the globe, with people across campus, and that was something that just didn't exist before?
Exactly, exactly. And we had students from all over the world in Kansas City, University of Missouri, Kansas City, and we had these students who would, as soon as they get a break during lunch or between classes, they will rush to the computer, you know, all these computer systems and log in feverishly.
And then they will get into this thing called IRC chat, Internet relay chat, which is kind of like this, the equivalent of Instagram or WhatsApp or whatever, whatnot.
So they would be chatting with their friends all across the globe.
So it was very, very fun to watch how people connected those days.
And it's nothing different from how they connected today. It's just the mechanisms were different, but people were so enthusiastic about being able to talk to their friends and families back home and connect to all their friends in the U.S.
So it brought everybody together. It was the first feeling of having a sort of like a global village where we can all connect no matter where we are dispersed across the globe.
Yeah, that kind of global town hall. I know in the IRC era, I was pleasantly surprised.
My small town, one of the kids set up an IRC room.
So what you found was those that had a computer and could get in that room, you'd hang out at school all day.
And then after school, everyone would jump on their computers and continue that social experience of high school through an IRC chat room.
And you'd still chat with the people and you could interact with kids older and younger, which was just an interesting experience that IRC and the Internet allowed.
Exactly, exactly. Those were the main applications, emails, IRC chats.
And also we had file transfer protocol, which is basically FTP, as we fondly call it, where you can drop and pick up big blocks of files, source code, whatever, we can share that using FTP.
So if somebody wants to upload a software, they would just open up a FTP server, give their customers a guest username and password, and they will drop in and pick up the code or whatever.
So FTP was another protocol and mechanism to exchange big blocks of files those days.
And did you find that, what was, I guess, early computer science education like when, you know, this is your first instance with email, you're still working on Sun Microsystems machines, what was the beginning of that educational process?
So, interesting question, Dan.
So, when I started my MS program, I asked my cousin, he was in Digital Equipment Corporation, and asked him, what is the program should I focus on?
And he told me, hey, I feel this whole telecom deregulation and this whole telecom revolution is about to happen.
So there is this interesting program in University of Missouri, Kansas City, where the computer science program is just not about programming, it's more about telecommunications and Internetworking.
So why don't you consider that and specialize there, because you would get a fast start in this industry, which is going to explode.
And he was exactly on the mark.
So I signed up for all the telecom courses and networking courses.
So I actually got a MS in computer science and telecom program on Internetworking speciality.
So this was almost the, like, for those that have seen the graduate, this is almost the plastics, kind of, like, what should I study?
Oh, just get into telecom, trust me, like, you'll do fine.
You'll do fine. So I just showed up and started taking all these communication courses and my professor started teaching me fiber optic communications, circuit switching versus packet switching.
So I had a very early interest and very early place to learn Internetworking, even in college, which was a very unique program at that time.
And really helped me in my career from that point onwards.
So when I came to the Silicon Valley in 93, once again, I applied to a networking company called the Wollongong Group.
What Wollongong Group was doing was they were building a TCP stack for the Windows 3.1 operating system on the desktop.
It did not have a native TCP stack built in.
Imagine an operating system without a networking stack. Those were the days, right?
So Wollongong Group was selling a product called the TCP Runtime, where basically you would load the TCP stack onto a Windows 95 machine, and then you will run applications like email and FTP and Telnet that I spoke to you about.
But this is for commercial use for enterprises and colleges and educational universities.
So Wollongong was selling the stack and also the applications on top of the stack for PC, DOS operating systems.
So I was able to carry my coursework and the knowledge that I gained in university into real world.
And I started, lo and behold, I started testing their email application once again in Wollongong Group.
So I was an email test engineer. And that's fascinating to think, you know, what is just built in today to an operating system into the technology we use.
You know, there was an era in time where, no, it didn't come with the networking software or the drivers because, you know, PCs weren't shipping with modems back then.
Once you got the hardware, yeah, the operating system didn't need to care at that stage about TCIP connections.
And so it's interesting to see someone then fill that niche or a company come in and say, well, yeah, exactly.
Still need to build it. Yep. So that's where I was for a couple of years. And then I got really enamored with this company called Cisco Systems.
They were making these devices.
No one's heard of them. You might have to go into more detail.
Right. So I started as a very entry level support engineer in Cisco in January of 96.
So that was my foray into the hardware and the infrastructure and the routing portions of the Internet.
So I went from the PC TCP stack on the DOS, which is the end system to a networking device, which is like a router, which is basically an intermediate system connecting all these disparate networks together.
Those days we had very many different networks, Dan. We had Token Ring.
We have IBM protocols, which was running in mainframe networking. And then we had Ethernet and we had Novel Network.
We had all these type of hardware and software technologies where you have to connect all these disparate networks together.
And that's where Cisco made its mark. It was able to connect AppleTalk, which is a protocol for Apple computers, and then Novel Network for PCs, and then IP routing for connecting the Internet together.
So it had this multi-protocol operating system, which could route packets between these different networks.
And that was basically their solution in connecting all these different networks in universities and educational institutions and commercial enterprises.
So I started as a support engineer, supporting their switches and routers in 96.
And it had exponential growth because right around the same time, Mark Anderson from University of Illinois Urbana-Champaign had this cool application called Web Browser, where we can go and download this web browser called Mosaic for free from University of Illinois Urbana-Champaign and put it on your Windows computer or in your Unix computer, and you start browsing all these websites.
So there were only a handful of websites during those times, but that was the killer application that the Internet was waiting for, where the bandwidth usage kind of exploded, 5x, 10x, in a very short period of time.
So in order to keep up with that, Cisco had to invest in optical networking technologies to drive more packets and more bandwidth, because the network was growing 5x, 10x every other year.
So to keep up with the demand, we were going from T1 circuits, which was like 24x64, like 1024 type packet links to almost DS3, which is 45 megabits per second, from there to OC3, which is 155 megabits per second, and then OC192, which is 10 gigabits worth of throughput to the Internet.
And then so on and so on. So we were going from basically a T1 line, which is basically 24x64 kilobits per second to almost like 10 gig to the Internet.
So in a span of four years. And I guess, was this growth surprising to you?
Did it just make sense from your own personal experiences on the Internet?
And you knew like, yep, like a highway, the more bandwidth you get, the more it's just going to be used.
And, you know, even looking at today's usage of the Internet, which is very video heavy, very bandwidth heavy.
And, you know, the moment, if 4k and 8k are the next bastions of streaming technology, that just requires bigger pipes.
Bigger pipes, correct. So once we saw the same time, there were so many dot com companies that came into picture, amazon.com, etoys.com, and they were all having all of this.
Everybody wanted to be on the Internet.
Everybody wanted to sell something on the Internet. So the companies like Amazon, it gave birth to companies like amazon.com, right?
And then everything was going fine until we had this Y2K problem where you have to fix the computer so that when the year 2000 rolls on and we hit zero zero, the computers don't roll over and die, right?
So we had this huge upgrade cycle in hardware and software to patch the Y2K bug.
So we pulled in so much demand from the future to address this one bug.
And all of a sudden after we fixed that Y2K bug and, you know, Y2K came and went, none of the networks died, none of the computers died.
And then there was a huge pause in demand at the tail end of 2000.
So we started seeing a huge decline in demand of networking equipment, computers, and everything.
And then lo and behold, the economy was slowing down. So we had this huge correction in stock market called the dot com burst in 2000.
And then we went through, while the rest of the world went into a, what do you say, a recession?
Telecom went into a depression because the whole investment by some of these companies with the heavily leveraged balance sheets caused some of them to go bankrupt.
Especially these optical fiber technology companies, they were laying out more and more optical fibers underneath and they couldn't sell them fast enough because the demand, they had overbuilt the infrastructure and they had to drop the prices.
So there was this huge churn in the telecom market, but actually what it spurred is that things gotten very cheap.
And all of a sudden there were so many computers out there in eBay.
So practically Google bought these servers for 10 cents on a dollar from eBay and built their network.
And they bought these Juniper routers, which would cost $28,000, $30,000 to buy for $2,000 in eBay.
So the entire Google network was built with this eBay solutions that they bought for one -tenth of the cost.
So that spurred a big innovation at one-tenth of the cost for Google.
So always look for a fire sale, I guess is what I'm hearing. Exactly. And so to be able to take then those disparate machines, those disparate servers, build what was early Google, it's a fascinating tidbit.
Yeah. So I was the support engineer.
Basically, I was the solutions engineer for Juniper, visiting Google as my customer.
When I started as a sales solutions engineer in Juniper, that was my first account that was given to me in 2004, 2005 timeframe.
When I walked into there, Google asked me, hey, Chilian, we had these routers we bought at eBay.
By the way, could you give us support?
I'm like, no, you bought it at eBay. Sorry, customer, you need to buy the new products, the latest products and buy support from me, unfortunately.
So they said, yeah, makes sense. I mean, those routers were getting old anyway.
So tell me about this new TC640 routers you have in the backbone.
We would like to buy a few of those. And me and my account manager walked in and asked them, Ray Testa and myself, he's the account manager, I'm the SCV.
We walked into the account. There was a gentleman there, supply chain manager.
He said, hey, Chilian, what happened to the previous account team?
We said, oh, sorry, Ike. They took a different account. They went with a different account.
We are the account team here. And they said, oh, really?
Aren't you the luckiest ones on the planet? We were like, why? Well, we had this new application called Google Maps, a company we bought called Keyhole Technologies.
All of a sudden, our bandwidth usage has exploded. We are running at 90% of our circuits.
We're going to upgrade this network. And here you are.
What was your quota? And we were like, we quoted a small quota. And they're like, oh, man, you're going to blow it out by 10x.
I'm going to write you a check. You have a different problem now, Chilian, because you have to supply all this equipment.
It's a huge deal, Dan. Imagine going from 1 million to 10 million, 10x. So they said, well, you have to go find equipment for us.
And we said, sorry, Ike, you came in at the last minute and you're asking me to take equipment from some other vendor and some other customer and give it to you.
And they said, well, you know what?
We will have Eric Schmidt call your CEO, Scott Cranes. We are done talking to you.
We'll just go to the top of the executive ladder. Yeah, exactly. You can offend a different customer.
We'll make this work. We'll make this work. So lo and behold, Eric got on the phone on the same day evening and called Scott Cranes.
And we got a call internally. And the team in Juniper called us and said, well, what happened, dudes?
You both showed up and all of a sudden there is a deal.
And then I got a call from Eric. What's going on? We said, oh, Scott, this is a nice to have problem.
We have a 10x growth. We need to sell them more routers. He said, okay, boys, we cannot give you all the routers.
We will give them 25 to 50% of that and we will straddle it over the next three to six months.
I will manage it.
So you boys just hold the fort. And then that's when we sold this huge deal to Google and started building out the Google fabric as well, which is a huge search engine fabric for the Internet.
And then, yeah, go ahead, Dan. Yeah. So was it in that moment that you were thinking back like, man, good thing I went into telecom?
Yeah, exactly.
Good thing I went into telecom. It was a big soul fulfilling thing for me because I saw the downturn in 2000 and I was questioning my life choices in career.
And then five years later, Google makes this huge offer, which kind of reinforced my faith in the Internet.
And I felt like, okay, there is demand for this type of equipment and there is demand for the growth of Internet again.
So that was the Web 2.0 companies, the birth of Google's and the Facebook's, the Web 2.0 companies.
And they had a different business model, as we all know.
They amortize the cost of the infrastructure and the application by serving the ads.
So they had a different tack in getting the money out of the business, which is pretty cool.
And then Google had this question about, hey, we have these routers, which sit in the middle of the network, and we have this code and servers and all of that.
These servers are able to adjust to our programming concepts much better than the routers.
And the routers are very static devices. They have this arcane command line interface where everybody has to log in and fetch these commands.
It's not scaling and it's not dynamic to the cloud needs.
So we want to make it very programmable, and we want to separate the control plane and the data plane, which is the hardware and the software, and have a software controller sitting in the cloud managing all these devices at the edges of the network and also in the core.
And that was the birth of software-defined networking, SDN, as you might know it.
And in a way, I look at the Cloudflare network, we are using SDN without even calling it SDN.
We have a controller in PDX, which is basically our Quicksilver database where we go and punch in the customer data or the customer account and things like that.
And then the Quicksilver database propagates it all over the edge. And our edge servers used to be just L7 proxy.
Now they're like a software router. Basically, they route traffic from the Internet onto a GRE tunnel where we are handing off traffic to an enterprise customer who's using Magic Transit as a transit service and advertising their networks, Layer 3 prefixes to the Internet.
And we are offering them a conduit pipe from the Internet to their premise.
And we have basically turned our L7 proxy edge servers into edge routers. We are routing traffic.
So it is an example of how we have taken the software -defined networking to the next level in having a controller and separating the control plane and the data plane, and then programming those devices at lightning speed.
So Cloudflare is an example of using SDN without even calling SDN. Yeah. And with that in mind, we're nearing the end of time.
What are you most excited about for the future of the Internet that you get to work on day to day here at Cloudflare or what's going on in the broader world?
Yeah. So Cloudflare sits in what you call in the old days AS701, which is the EUnet BGP autonomous system, used to be the center of the Internet.
And it's kind of flipping where Cloudflare and Google and Facebook, they are the center of the Internet.
And Cloudflare, with its traffic profile at the same level as Google and Facebook, becomes the core piece of the Internet and becomes a gateway for our enterprise customers, for our small and medium-sized customers and mid-market customers to connect to the Internet.
So I can see Cloudflare evolving more like an ISP or Internet service provider moving forward, offering these transit services and becoming really like the AS701, which used to be the glue of the Internet.
The Cloudflare AS, I can see becoming the glue of the Internet for a significant portion of the Internet.
And that is very exciting for us to be there.
I mean, think of that, right? 25 years ago, EUnet used to be the center of the Internet and Cloudflare has this great opportunity to be a center of the Internet.
Yeah, definitely a fascinating opportunity ahead of us and a big responsibility along those lines.
Yes, with great power comes great responsibility, right?
So we have all this great, so much traffic on us and using Cloudflare as an intermediate gateway to the Internet.
So we have a lot of responsibility in terms of handling the property, the customer data and the data regulations, privacy, all of those things puts enormous responsibility on our shoulders.
And we are doing a phenomenal job in managing that. So I think we have a very bright future, given the fact that we have gone into a market like Magic Transit and effectively we have a quadrupled that market in one year.
And moving forward, we are really excited about the growth of Magic Transit and then the transit services and becoming the center point or a touch point to the Internet for majority of our customers, no matter what their size and scale is.
They could be a $200 website or a $200,000 e-commerce site, doesn't matter.
We believe in the democratization of the Internet.
We are bringing services to everybody, building a safer, better Internet, and it kind of fits into our mission that we keep doing it.
And I'm very excited about that portion of Cloudflare. Do you almost see kind of an analogy to Cisco and how they were able to connect the various protocols, various systems, it really didn't matter size of company or spend, they found a way to do it cost effectively and still connect all of those devices and protocols?
Correct. Matthew Prince, when I joined the first time, when I saw him present to us in the orientation in 2018, told us that if I were to explain Cloudflare, in a few words, it is Cisco as a service.
That's what Matthew Prince said.
So it's kind of, it's correctly, aptly fits your description, Dan. We are Cisco as a service.
So here's to hopefully making that dream come true over these next few years.
So as we near the end of time, Shailen, I want to thank you for that trip down memory lane through the early Internet.
And that was just a fascinating story of your time at Google and before.
And it all started from an email address, essentially.
It all started from an email address. Thank you, Dan. Thanks for the wonderful opportunity to talk to a broader audience in Cloudflare.
Hope we can keep doing this more and more.
All right. Thank you, Shailen. And for those catching the live stream or recording, thank you very much for joining us and have a wonderful morning, afternoon or evening.
Thank you.
Transcribed by https://otter.ai