CTOcast # 3: Conversation with Sergey Chernyshev
Introducing the third issue of the podcast on technology, processes, infrastructure, and people in IT companies. Today CTOcast is visited by Sergey Chernyshev, one of the leaders of the Web Performance Optimization community (New York).
Listen to the podcast
About our interlocutor:
Sergey Chernyshev graduated from I.M. Gubkin Russian State University of Oil and Gas (2000). In the first half of the 1990s, he was engaged in computer assembly, system administration and technical support. In 1996–1998, he was a web developer at Unicor, and in 1998–2002, CTO at Novikov’s design studio. Since 2002 lives and works in the USA. In 2002–2003, he was a system administrator at SelectMinds. Since 2003 he has been working in truTV (formerly Court TV), since 2012 he has been the technical director of web systems and applications.
Since 2009, he has been actively involved in the Web Performance Optimization community. Organizer of the New York Web Performance Meetup Group, as well as WebPerfDays New York. Member of the Board of Web Performance Optimization Foundation.
Alexander Astapenko: What was the IT market in the early and mid-1990s when you were just starting your career? How did you manage to survive and what were your goals at this stage?
Sergei Chernyshev: From the very beginning, I was just looking for work. I can’t say that I immediately thought about my own business or something like that. A friend brought me to the usual computer collection company Tower, and just there I realized that any computer can be disassembled, assembled and not be afraid of a mythical machine. It also came to the realization that business is not so scary and everyone can do it. Soon, I teamed up with a couple of my friends, and we organized a service center where we helped people live with computers: collected what they needed, sold hardware, and so on.
At that time it was quite a fun activity. Several major suppliers carried containers of iron to Moscow, and a bunch of small companies sold them. Naturally, there could be no question of any guarantees or quality, each for itself. I had to follow this very much, and the experience gained in the Tower helped determine the quality of the components in appearance, even without inclusion.
I remember at that time everyone used such a directory in a finger thick. In my opinion, there were all kinds of goods in it, but first of all - computer ones. You could buy anything from that catalog. As a result, they did this: if there was an order, they specified what a person needed, took money and then bought and collected the necessary equipment, delivered it. The main thing that distinguished us was our quality - the quality of services. Each printer came with a bundle of paper, a person did not leave until everything worked.
Pavel Pavlov: However, at some point you started to switch to web development?
Sergey Chernyshev:Yes Yes. There were several reasons for this. Firstly, business needs to be done seriously, investments are needed, and our approach was quite simple, and for a long time this kind of business could not exist. In addition, my father was very active in telling me that a mechanical engineer is not a career, and you need to do more things in life. I probably did not listen to him, but I think that this nevertheless added serious thoughts to my head.
I started using the Internet around 1993, or even 1992, the year. At that time, the Web as such was not there. There was such an Internet provider Glasnet, one of the first in Moscow, and I actively participated in the local user community. Andrei Sebrant, a well-known Russian Internet activist, was a man who showed me and my colleagues what the Web is.
I was very impressed with this issue and started my development, as they say, on the knee back in the year 1994 - when TCP / IP just started appearing, you could start the first browser and look at something on the Internet. As I remember now, one of my first sites was Zhvanetsky’s magazine, and it was called Zhvanetsky’s Store. Of course, finding it now on the Internet is no longer possible.
Alexander Astapenko: Sergey, what was this story with Andrei Sebrant?
Sergey Chernyshev:Andrei Sebrant at that time was, in my opinion, the head of the marketing department at Glasnet. Although I doubt that then there were such clear distinctions. At one party of this Internet community, we were at Andrei’s house, and for the first time I saw what the Web is, what browsers are, I saw GIF images on the page. It just hit me. I think then I realized that I would deal with the Web. And then Andrei showed several tools. There was such a HotDog editor that allowed me to see what HTML is and how it generally works, how to start making the first pages.
Alexander Astapenko: And what was the development in the late 1990s? Who were the customers for whom you then made sites?
Sergey Chernyshev:At first I did it for friends and acquaintances. Then, while studying at Kerosink, I went to work as a web developer and system administrator for an Internet provider that provided Internet for the entire building. At that time there was no separation in professions, we were all called webmasters. The webmaster was engaged in design, HTML, server coding, hosting and everything else.
Later I went to Sergey Novikov, who was engaged in design and wanted to create a web design studio, and joined him as the technical co-founder of the company. We have made websites for everyone who wishes. There were many companies, they often came with their vision of what they needed. The main work was to convince them, because most often then people had little idea of what websites were. At that time, there was no such WordPress to take and build something on it, everything was written from scratch.
Alexander Astapenko: Clearly, there was such a wild West on the Web then ... Around 2002, you moved to the USA. Can you say a few words about this period?
Sergey Chernyshev:Yes. Immigration is a complex issue; it cannot be said that there was only one incentive. I wanted to become rich and happy, and looking abroad foreshadowed such prospects. Well, and family, you had to take care.
Alexander Astapenko: And from a professional point of view?
Sergei Chernyshev: From a professional point of view, it was clear that business in Russia is very corrupt and it was hard to continue to do this. In principle, moving forward required indiscrimination in how you conduct business. Many managed to do this, but I didn’t. And in America, of course, there was a great desire to join the online get-together and see the vast possibilities.
Alexander Astapenko:Sergey, 2002 - the very bottom of the crisis after the collapse of the dotcoms. What then was the market for a Russian web developer, a system administrator with several years of experience behind? Was it easy to find a job?
Sergei Chernyshev: Yes, just in 2002, it was the bottom, really. From the point of view of moving, I’ll say this: you plan to move long enough and hard to know that everything will be bad, that is really happening. The car starts to move, and you move when everything has grown together. Therefore, it was not necessary to plan especially. It may be easier in our time, but at that time it was.
Arriving here, I realized that the market is bad enough, however, I managed to find a job quickly. The first principle was that first place is not chosen. After three or four interviews, maybe five, I already found SelectMinds. True, from the head of the studio in Russia, I just switched to a system administrator in a small company. When moving, the work history was reset to zero, no one considered foreign experience for various reasons: the business is different, you cannot verify and trust, nobody can check the resume. And, in general, only the skills that you could show, only they influenced.
Alexander Astapenko: After working at SelectMinds as a system administrator, you still return to web development and transfer to Court TV. So?
Sergey Chernyshev:Yes. At that time, finding a job was already much easier and I was interested in Court TV. The first couple of years was not difficult, although the traffic was significant due to the popularity of the channel and its website. After that, more complex projects began.
We were one of the first to start broadcasting on the Web. Court TV began to cover criminal stories and trials of criminals. We've got a few of these court cases like “The Court at Michael Jackson,” or there was such a high-profile Scott Peterson case. At the time of the announcement of the verdict, traffic on our servers increased twenty times compared with the maximum on a normal day.
And this made me take up significant architecture for scalability, understand how to build such systems and develop development in this direction. Which, in the end, led me to web performance (web performance): because where there are problems with scalability, there will be problems with performance.
Pavel Pavlov: Could you explain for those who are not quite familiar with this topic, what is web performance optimization? How did this movement originate? You were at the beginning, 2009 is the very beginning?
Sergei Chernyshev: In fact, since 2007 I began to actively engage in this. I attended the first presentation by Steve Souders - who was the founder of the WPO (web performance optimization) movement and who worked at Yahoo at the time - on the issue of website speed. At that time, the term web performance optimization did not even exist.
I think Steve opened his eyes to many. Everyone understood that the slower the speed, the worse the website. There is no need to go far. But most people have traditionally associated site speed with server speed. The usual task was to combat traffic: when a lot of people come and they require a lot of server resources, scalability is a serious problem and you need to make sure that websites continue to work. This was the focus for developers, that is, it was considered speed.
And Steve showed that the speed depends more on how the web page is built and what it loads, the front end, and not on the load on the system. A website can be slow even for a single user. Therefore, you need to look for the problem elsewhere.
And at that time, of course, there were no tools, or rather, there were very simple tools, it was quite difficult to use them and no one knew about them. Therefore, this opacity of the web stack stopped people, they could not see the real problem. And Steve showed that if you look from the front end, from the front end, on how the HTTP protocol works, then you can draw much more conclusions about the real speed of the website.
I have been doing gzip for a long time, compressing web requests, cache control and so on. It was clear to me what the problem was. And the first presentation that Steve Soders made at the Web 2.0 Expo in San Francisco opened my eyes, and I regained my sight, as they say.
Pavel Pavlov: And you personally attended there?
Sergey Chernyshev:Yes. I came to this conference as a visitor. Steve and his colleague (Tenni Theurer) presented excellent research that they did at Yahoo regarding statistics, analyzed what really affects the speed of sites, how the front-end part affects. The very first rules that Steve introduced to speed up hardware-free websites that are server-independent.
Pavel Pavlov: Did you manage to talk to Steve himself then?
Sergey Chernyshev:Yes, I talked to him, met at this conference. It was very interesting. I think that he did inspire me to create a mitap. At the same conference, but already in 2009, I asked him how he relates to such an idea. Conferences were the right approach to disseminating knowledge, but while in New York, it was very difficult to wait a whole year before the next conference. I was already looking for ways to create local events. Steve replied that it would be interesting to see some local community. In general, this spurred my interest in creating the first New York metap.
Pavel Pavlov: It turns out that the mitap in New York was one of the first in the history of group rallies for web performance?
Sergey Chernyshev:He was the first literally. When I created my group, I created meetup.com and the Web Performance topic, it did not exist before. And now there are already more than 100 groups on this subject from around the world.
Pavel Pavlov: Who were the first participants? And what was the reaction of people when you tried to promote all these topics, to tell?
Sergey Chernyshev:Before opening my mitap, I made a performance presentation at Web Standards Meetup in New York. It was a small community, at that presentation there were about 35–40 people. And people opened their eyes, as I did when I saw what Steve Soders was working with, his approaches. People simply had no idea how the browser loads images, scripts, and CSS files. For them it was not obvious how this all happens, no one thought about it. And as soon as you visualize and show the first waterfall diagram, people begin to understand. At that time, browsers worked quite poorly and it is easy to imagine what this waterfall looked like - everything was very slow. Therefore, it was easy to show what the performance problem was.
A maximum of 20% of the speed depends on the server, but in reality the figure is closer to 10% or even 5%. Everything else depends on how the web page is made, in what order its components are loaded. And people were really stunned by how everything happens in reality. This showed me that the question is relevant.
Pavel Pavlov: Let's go back to the origins of WPO and to your meeting with Steve. In addition to creating the first WPO-meeting in New York, he also inspired you to develop your own tools and solutions that would help make websites faster and allow you to evaluate their speed. So?
Sergey Chernyshev:That's right. As I said, the problem of web performance was the lack of tools and the opacity of the stack. Steve was one of the first developers of the tool from Yahoo - YSlow, which analyzed websites to support best practices and simply gave out some metrics, telling the person that his site was so good (from 0 to 100) and what can be done, to improve it.
The problem with this tool was that it did not provide a long-term analysis of how development changes occur. Web performance tends to deteriorate, that is, with the development of new parts of programs, new features, the quality of programs in terms of speed only deteriorates. With a one-time static analysis, it is very difficult to notice. I saw that this was a problem, and therefore I developed the first version of Show Slow, which allowed collecting data from the extension for Firebug - YSlow. And I created a server component that just accepted this data. The program is nothing complicated, but it collects data and allows you to identify a trend over time.
As it turned out, this significantly helped to show how development based on speed is important for the business. I came across this at truTV when it was very difficult to abstractly tell the business, both technical people and customers, that it was necessary to speed up the website. It’s easy to say, but no one believes you, it was not clear to people why this should be done, all the more seriously, to spend a lot of time. And Show Slow was created as the first tool that allows not-so-technical people to see the result without using sophisticated tools like YSlow to show the trend of how speed improvement or deterioration is happening.
Naturally, in addition to tools, I also began to engage in various methods and ways to speed up websites. And the cache was the first way. The easiest method was to use the cache so that the browser does not go back to the server every time for all the pictures and so on. And I developed Asset Manager, which, roughly speaking, used revisions from the version control system and thus manipulated the URL. After that I created a .htaccess file, which is very easy to put in the root of the website, and it magically accelerated the entire website, which was a slightly marketing move. I chose this path to convince those people who are not very technical that it is easy to solve the speed problem. And, in my opinion, many liked it. In my opinion, this is my most popular project at the moment on GitHub. People really want a quick and easy solution.
The text version of the podcast will continue in the coming days.
Listen to the podcastAbout our interlocutor:
Sergey Chernyshev graduated from I.M. Gubkin Russian State University of Oil and Gas (2000). In the first half of the 1990s, he was engaged in computer assembly, system administration and technical support. In 1996–1998, he was a web developer at Unicor, and in 1998–2002, CTO at Novikov’s design studio. Since 2002 lives and works in the USA. In 2002–2003, he was a system administrator at SelectMinds. Since 2003 he has been working in truTV (formerly Court TV), since 2012 he has been the technical director of web systems and applications.
Since 2009, he has been actively involved in the Web Performance Optimization community. Organizer of the New York Web Performance Meetup Group, as well as WebPerfDays New York. Member of the Board of Web Performance Optimization Foundation.
Text version of the podcast (1st part)
About personal and a little about the history of the Russian IT industry
Alexander Astapenko: What was the IT market in the early and mid-1990s when you were just starting your career? How did you manage to survive and what were your goals at this stage?
Sergei Chernyshev: From the very beginning, I was just looking for work. I can’t say that I immediately thought about my own business or something like that. A friend brought me to the usual computer collection company Tower, and just there I realized that any computer can be disassembled, assembled and not be afraid of a mythical machine. It also came to the realization that business is not so scary and everyone can do it. Soon, I teamed up with a couple of my friends, and we organized a service center where we helped people live with computers: collected what they needed, sold hardware, and so on.
At that time it was quite a fun activity. Several major suppliers carried containers of iron to Moscow, and a bunch of small companies sold them. Naturally, there could be no question of any guarantees or quality, each for itself. I had to follow this very much, and the experience gained in the Tower helped determine the quality of the components in appearance, even without inclusion.
I remember at that time everyone used such a directory in a finger thick. In my opinion, there were all kinds of goods in it, but first of all - computer ones. You could buy anything from that catalog. As a result, they did this: if there was an order, they specified what a person needed, took money and then bought and collected the necessary equipment, delivered it. The main thing that distinguished us was our quality - the quality of services. Each printer came with a bundle of paper, a person did not leave until everything worked.
Pavel Pavlov: However, at some point you started to switch to web development?
Sergey Chernyshev:Yes Yes. There were several reasons for this. Firstly, business needs to be done seriously, investments are needed, and our approach was quite simple, and for a long time this kind of business could not exist. In addition, my father was very active in telling me that a mechanical engineer is not a career, and you need to do more things in life. I probably did not listen to him, but I think that this nevertheless added serious thoughts to my head.
I started using the Internet around 1993, or even 1992, the year. At that time, the Web as such was not there. There was such an Internet provider Glasnet, one of the first in Moscow, and I actively participated in the local user community. Andrei Sebrant, a well-known Russian Internet activist, was a man who showed me and my colleagues what the Web is.
I was very impressed with this issue and started my development, as they say, on the knee back in the year 1994 - when TCP / IP just started appearing, you could start the first browser and look at something on the Internet. As I remember now, one of my first sites was Zhvanetsky’s magazine, and it was called Zhvanetsky’s Store. Of course, finding it now on the Internet is no longer possible.
Alexander Astapenko: Sergey, what was this story with Andrei Sebrant?
Sergey Chernyshev:Andrei Sebrant at that time was, in my opinion, the head of the marketing department at Glasnet. Although I doubt that then there were such clear distinctions. At one party of this Internet community, we were at Andrei’s house, and for the first time I saw what the Web is, what browsers are, I saw GIF images on the page. It just hit me. I think then I realized that I would deal with the Web. And then Andrei showed several tools. There was such a HotDog editor that allowed me to see what HTML is and how it generally works, how to start making the first pages.
Alexander Astapenko: And what was the development in the late 1990s? Who were the customers for whom you then made sites?
Sergey Chernyshev:At first I did it for friends and acquaintances. Then, while studying at Kerosink, I went to work as a web developer and system administrator for an Internet provider that provided Internet for the entire building. At that time there was no separation in professions, we were all called webmasters. The webmaster was engaged in design, HTML, server coding, hosting and everything else.
Later I went to Sergey Novikov, who was engaged in design and wanted to create a web design studio, and joined him as the technical co-founder of the company. We have made websites for everyone who wishes. There were many companies, they often came with their vision of what they needed. The main work was to convince them, because most often then people had little idea of what websites were. At that time, there was no such WordPress to take and build something on it, everything was written from scratch.
Alexander Astapenko: Clearly, there was such a wild West on the Web then ... Around 2002, you moved to the USA. Can you say a few words about this period?
Sergey Chernyshev:Yes. Immigration is a complex issue; it cannot be said that there was only one incentive. I wanted to become rich and happy, and looking abroad foreshadowed such prospects. Well, and family, you had to take care.
Alexander Astapenko: And from a professional point of view?
Sergei Chernyshev: From a professional point of view, it was clear that business in Russia is very corrupt and it was hard to continue to do this. In principle, moving forward required indiscrimination in how you conduct business. Many managed to do this, but I didn’t. And in America, of course, there was a great desire to join the online get-together and see the vast possibilities.
Alexander Astapenko:Sergey, 2002 - the very bottom of the crisis after the collapse of the dotcoms. What then was the market for a Russian web developer, a system administrator with several years of experience behind? Was it easy to find a job?
Sergei Chernyshev: Yes, just in 2002, it was the bottom, really. From the point of view of moving, I’ll say this: you plan to move long enough and hard to know that everything will be bad, that is really happening. The car starts to move, and you move when everything has grown together. Therefore, it was not necessary to plan especially. It may be easier in our time, but at that time it was.
Arriving here, I realized that the market is bad enough, however, I managed to find a job quickly. The first principle was that first place is not chosen. After three or four interviews, maybe five, I already found SelectMinds. True, from the head of the studio in Russia, I just switched to a system administrator in a small company. When moving, the work history was reset to zero, no one considered foreign experience for various reasons: the business is different, you cannot verify and trust, nobody can check the resume. And, in general, only the skills that you could show, only they influenced.
Alexander Astapenko: After working at SelectMinds as a system administrator, you still return to web development and transfer to Court TV. So?
Sergey Chernyshev:Yes. At that time, finding a job was already much easier and I was interested in Court TV. The first couple of years was not difficult, although the traffic was significant due to the popularity of the channel and its website. After that, more complex projects began.
We were one of the first to start broadcasting on the Web. Court TV began to cover criminal stories and trials of criminals. We've got a few of these court cases like “The Court at Michael Jackson,” or there was such a high-profile Scott Peterson case. At the time of the announcement of the verdict, traffic on our servers increased twenty times compared with the maximum on a normal day.
And this made me take up significant architecture for scalability, understand how to build such systems and develop development in this direction. Which, in the end, led me to web performance (web performance): because where there are problems with scalability, there will be problems with performance.
About optimizing web performance
Pavel Pavlov: Could you explain for those who are not quite familiar with this topic, what is web performance optimization? How did this movement originate? You were at the beginning, 2009 is the very beginning?
Sergei Chernyshev: In fact, since 2007 I began to actively engage in this. I attended the first presentation by Steve Souders - who was the founder of the WPO (web performance optimization) movement and who worked at Yahoo at the time - on the issue of website speed. At that time, the term web performance optimization did not even exist.
I think Steve opened his eyes to many. Everyone understood that the slower the speed, the worse the website. There is no need to go far. But most people have traditionally associated site speed with server speed. The usual task was to combat traffic: when a lot of people come and they require a lot of server resources, scalability is a serious problem and you need to make sure that websites continue to work. This was the focus for developers, that is, it was considered speed.
And Steve showed that the speed depends more on how the web page is built and what it loads, the front end, and not on the load on the system. A website can be slow even for a single user. Therefore, you need to look for the problem elsewhere.
And at that time, of course, there were no tools, or rather, there were very simple tools, it was quite difficult to use them and no one knew about them. Therefore, this opacity of the web stack stopped people, they could not see the real problem. And Steve showed that if you look from the front end, from the front end, on how the HTTP protocol works, then you can draw much more conclusions about the real speed of the website.
I have been doing gzip for a long time, compressing web requests, cache control and so on. It was clear to me what the problem was. And the first presentation that Steve Soders made at the Web 2.0 Expo in San Francisco opened my eyes, and I regained my sight, as they say.
Pavel Pavlov: And you personally attended there?
Sergey Chernyshev:Yes. I came to this conference as a visitor. Steve and his colleague (Tenni Theurer) presented excellent research that they did at Yahoo regarding statistics, analyzed what really affects the speed of sites, how the front-end part affects. The very first rules that Steve introduced to speed up hardware-free websites that are server-independent.
Pavel Pavlov: Did you manage to talk to Steve himself then?
Sergey Chernyshev:Yes, I talked to him, met at this conference. It was very interesting. I think that he did inspire me to create a mitap. At the same conference, but already in 2009, I asked him how he relates to such an idea. Conferences were the right approach to disseminating knowledge, but while in New York, it was very difficult to wait a whole year before the next conference. I was already looking for ways to create local events. Steve replied that it would be interesting to see some local community. In general, this spurred my interest in creating the first New York metap.
Pavel Pavlov: It turns out that the mitap in New York was one of the first in the history of group rallies for web performance?
Sergey Chernyshev:He was the first literally. When I created my group, I created meetup.com and the Web Performance topic, it did not exist before. And now there are already more than 100 groups on this subject from around the world.
Pavel Pavlov: Who were the first participants? And what was the reaction of people when you tried to promote all these topics, to tell?
Sergey Chernyshev:Before opening my mitap, I made a performance presentation at Web Standards Meetup in New York. It was a small community, at that presentation there were about 35–40 people. And people opened their eyes, as I did when I saw what Steve Soders was working with, his approaches. People simply had no idea how the browser loads images, scripts, and CSS files. For them it was not obvious how this all happens, no one thought about it. And as soon as you visualize and show the first waterfall diagram, people begin to understand. At that time, browsers worked quite poorly and it is easy to imagine what this waterfall looked like - everything was very slow. Therefore, it was easy to show what the performance problem was.
A maximum of 20% of the speed depends on the server, but in reality the figure is closer to 10% or even 5%. Everything else depends on how the web page is made, in what order its components are loaded. And people were really stunned by how everything happens in reality. This showed me that the question is relevant.
Pavel Pavlov: Let's go back to the origins of WPO and to your meeting with Steve. In addition to creating the first WPO-meeting in New York, he also inspired you to develop your own tools and solutions that would help make websites faster and allow you to evaluate their speed. So?
Sergey Chernyshev:That's right. As I said, the problem of web performance was the lack of tools and the opacity of the stack. Steve was one of the first developers of the tool from Yahoo - YSlow, which analyzed websites to support best practices and simply gave out some metrics, telling the person that his site was so good (from 0 to 100) and what can be done, to improve it.
The problem with this tool was that it did not provide a long-term analysis of how development changes occur. Web performance tends to deteriorate, that is, with the development of new parts of programs, new features, the quality of programs in terms of speed only deteriorates. With a one-time static analysis, it is very difficult to notice. I saw that this was a problem, and therefore I developed the first version of Show Slow, which allowed collecting data from the extension for Firebug - YSlow. And I created a server component that just accepted this data. The program is nothing complicated, but it collects data and allows you to identify a trend over time.
As it turned out, this significantly helped to show how development based on speed is important for the business. I came across this at truTV when it was very difficult to abstractly tell the business, both technical people and customers, that it was necessary to speed up the website. It’s easy to say, but no one believes you, it was not clear to people why this should be done, all the more seriously, to spend a lot of time. And Show Slow was created as the first tool that allows not-so-technical people to see the result without using sophisticated tools like YSlow to show the trend of how speed improvement or deterioration is happening.
Naturally, in addition to tools, I also began to engage in various methods and ways to speed up websites. And the cache was the first way. The easiest method was to use the cache so that the browser does not go back to the server every time for all the pictures and so on. And I developed Asset Manager, which, roughly speaking, used revisions from the version control system and thus manipulated the URL. After that I created a .htaccess file, which is very easy to put in the root of the website, and it magically accelerated the entire website, which was a slightly marketing move. I chose this path to convince those people who are not very technical that it is easy to solve the speed problem. And, in my opinion, many liked it. In my opinion, this is my most popular project at the moment on GitHub. People really want a quick and easy solution.
The text version of the podcast will continue in the coming days.