LinkedIn VP of Technical Operations: Lloyd Taylor

Dan Farber: Lloyd, thanks for joining me.

Lloyd Taylor: And thanks for inviting me.

Dan Farber: Lloyd you have been at LinkedIn since September of this year, what is LinkedIn and what compelled you to go there?

Lloyd Taylor: I think, LinkedIn is probably the next big thing in the Internet world. Why? The focus is how do we take executives and professionals and make them more effective. Anybody who has studied how good executives become effective, really notices two key things how well do they communicate, and how well do they locate and consume important information. So, that is the core mission of LinkedIn. It is to figure out how to help executives and professionals not only make the connections they need to be successful in their own job, but to leverage their knowledge and build their reputation in being able to help others. So, that's sort of the key. What is so fascinating about this is that it is a whole new way of thinking about essentially mapping what already exists in the real world, which is your network, my network, our information consuming, or looking for information into a highly leveraged environment. If I can make your job easier, if I can make you more successful through the tools that my company can provide, you are going to be a user for life and that is what makes a great company, right? It is high value, high perceived value and low perceived cost of using it. And that is the name of the game.

Dan Farber: The social network space is obviously very crowded and often confusing because everyone is kind of plumped in together, whether it is MySpace or Facebook or LinkedIn. How do you fit LinkedIn into this competitive space?

Lloyd Taylor: So, the way I like to think about it and I will probably get into trouble for saying it, but that is OK. Each of us has sort of different spheres of life. We have the work sphere, we have the home sphere. We have the social sphere. The way I like to think about it is we focus on the work sphere. LinkedIn is what you do when you are at work. And the other social networks are what you do at home, family and friends or maybe what you do at the bar, connecting with people finding out new with people or at the sports stadium. I think, there is absolutely a role for social networks that are based on my friends and social networks that are based on social outreach, dating and things like that. But, neither one of those really can be, in my opinion, morphed into a business focused network, because there is too much of the other idea wrapped into it.

Dan Farber: Do you think that all these various social networks should be able to integrate with one another in some fashion?

Lloyd Taylor: Or would you want them to?

Dan Farber: Well, to the degree that if I am on LinkedIn and I'd want to know, well, is this person on my LinkedIn list someone else who I have on my Facebook list and do those kids of interconnections?

Lloyd Taylor: I think it is a possibility. I think part of it’s understanding how people like to work, again focusing on what the end user wants. I think things like social, like family social and dating social networks. Do you really want to be doing the connection between that and your professional life? I think, there is risk in doing that. One is you have your professional persona. Does the fact that you happen to like old Marilyn Monroe movies is that relevant to your business or could it potentially harm your business perspective?

Dan Farber: And we have seen cases where prospective employers are looking at people's profiles on different social networks and deciding whether they want to hire people or not.

Lloyd Taylor: Exactly.

Dan Farber: So, it is certainly something people need to manage a lot better. LinkedIn is a fast growing network, 16 million 17 million members at this point, growing about a million per month, as I recall. And you are a datacenter guru. So, what are the datacenter considerations as you are scaling up this company?

Lloyd Taylor: I am really more of a scaling guru than a datacenter guru. It is just that lot of the interesting scaling problems over the last 10 years have been, how do we house these infinite numbers of computers. That's a solved problem now. I mean, it's boring, I am sorry. So, I have got to go and find other things that are interesting to scale. I was quite surprised when I joined the company. Usually, when I come into a small company, the infrastructure is burning to the ground, and that is the first thing I have to fix. Well, to the credit of those folks who are still there and have been there for three or four years, there is a rock solid infrastructure they designed and built. It is very uncommon to come into a company and see that well done. The challenge in scaling LinkedIn is really not the number of visits. I mean, this is a high value site as opposed to a super high traffic site. So, the big challenge there is the key value in any kind of social networking or business networking site is the graph or the connectivity map of how people are connected. It is a very complex graph theory problem. And the big challenge here for me and it is very different than my previous companies is not how do I handle a billion queries per second, it's how do I get the maximum value for my users out of their connectivity.

So, there is a lot of very interesting stuff going on in graph theory right now in Stanford, some at MIT. There’s actually some interesting quantum applications of graph theory. And a lot of what I am trying to do now is to understand how to add value, how to add capacity and capability to the graph or what we call the cloud, so that our end users can leverage the connections they already have.

Dan Farber: And leverage them at really work speed?

Lloyd Taylor: Exactly.

Dan Farber: And when you talk about rock solid infrastructure, coming from your background, what does rock solid mean to you?

Lloyd Taylor: The site is up all the time. And you know, the site doesn't go down because of hardware problems or network problems. You know, obviously, when you are pushing software out quickly and trying new things, you are going to have these issues with that. As I often say, if you don't fail, you are not pushing hard enough.

Dan Farber: Now you have spent three years at Google as the Director of Global Operations, building out datacenters. Obviously, that is very large scale. What are some of the issues that you faced in building out those datacenters, especially given what we have been talking about that it's the user experience that counts. I know that Google is very focused on speed, because if you can't deliver the results fast enough and accurately enough, people will go somewhere else.

Lloyd Taylor: So, the problem at Google, when I joined, was that it was no longer possible to get enough co location space to handle the growth of the company. The size of the computing infrastructure that Google uses is insanely large. A surprisingly small percentage of it is responsible for delivering end user services. Generally, you want to put that close to the end users. So, the idea of having smaller datacenters around the world to provide local access is one part of the strategy. But, Google's key value, in a lot of ways, is the quality of their algorithms how good the search results are. In order to do that, you have to throw insanely massive amounts of computing that you can run these very complex and very highly parallel algorithms on the web data. So, a lot of the challenge was how do we get enough space and how do we do it in an affordable way. Google's great secret and great source of success is the fact that everything goes back to basic physics there. If you want to think about how do you do web search, you go back to OK, what are people really trying to accomplish. So, what we did at Google is essentially throw out the book on how to build datacenters, and went back to basic heat transfer theory, went back to basic electrical theory. And essentially threw away everything that wasn't strictly necessary in sort of a minimalist design. And then figured out how to use..... Essentially the integrated circuit industry concept is step and repeat. So, instead of building chips on a silicon surface, what we figured out how to do was to buy 1000 acres in Oklahoma, and then build the datacenter. And when we need more space, actually be able to step and repeat that data center, build another one right next to it. And so that is the environment they have now. So, they have large numbers amounts of land in various places around the world. And as the business needs more space, they have a mechanism where they can simply start from scratch and have a new data center up and running in less than six months.

Dan Farber: In less than six months?

Lloyd Taylor: Yes.

Dan Farber: It is quite amazing! And then, in terms of the way all of the datacenters interact, if one goes down or there is some earthquake or whatever it might be, you obviously have ways in which that load can be transferred instantly?

Lloyd Taylor: Well it is really..... it isn't even instantly, it is that it never actually disappears. Part of being a good system design is understanding what level one wants to put the redundancy at. Again most of the CIOs are going to go with, 'let's have multiple sets of hardware'. Well another way to think about that is 'let us use the software layer or the network layer to pay attention to where things aren't working and redirect load'.

So, that is kind of the approach one has to take to be most efficient. It would be very expensive to replicate everything that Google does. On the other hand, it is not that hard to do N+1, N+2, and then in the software layer, keep track of what isn't working. And so, the end user would never see anything go down. Even if a meteor hits something in Oregon, you may have to hit reload to get your search, but that is all that is going to happen.

Dan Farber: Now, you have only been at LinkedIn for a few months. Going forward, what do you see is kind of the big challenge that you have to deal with there?

Lloyd Taylor: The number one challenge is getting value out of the mesh or out of the cloud for the end user. So, if we can make it easy for someone like yourself, let's say you want to interview somebody on a given topic, what we would love to have is that you, as Dan, can go and log in and say, "Hey I want somebody who is within 20 miles of my studio, who is an expert on XYZ." And immediately come back to you with high value of responses. You can also allow people in your network to say I'd like to be contacted for this sort of this thing. So, if we can cut through a lot of the challenge of how do I find somebody who is willing to talk to me, or how do I find somebody who has a certain skill, that is essentially mining the value of the network, and that's the big exciting thing that I believe is happening in the coming year. As there is a lot of fascinating stuff going on inside right now, certainly can't about it yet, but in the next few months, there is going to be some fundamentally exciting things that come out about how do we help people like yourself, like myself, anybody who needs to find sources of good information. What do executives do? They communicate and they consume information. If I can help an executive consume information that is most useful to them, my value goes up. If I can help them communicate with people who can be most helpful to them in achieving their goals, the value goes up.

Dan Farber: Now, it looks like you have spent an entire career innovating, how do you create that culture of innovation in these different places that you go, how do you inspire people to kind of break all the rules?

Lloyd Taylor: The culture needs to reward failure. That is the answer.

Dan Farber: Thanks for joining me today

Lloyd Taylor: I’m delighted to be here.

Dan Farber: I have been speaking with Lloyd Taylor who is the Vice President of Technical Operations at LinkedIn. For CIO Sessions, I'm Dan Farber. Thanks for watching.