Q&A: Is Your Data Center ‘Distracting’ You?
Steve Rubinow knows what business he’s in, and it’s not running data centers. He the CTO of Catalina, the digital media company that sends personalized offers and coupons to 260 million shoppers.
Transform to Better Perform asked him about some recent changes in his company’s IT structure and the impact that has had on its business.
We've got to be able to change what we do quickly because the time frames in which people measure things are getting smaller and smaller. That's the nature of the digital world.
Transform: Steve, can you tell me about what you've done with your systems in response to the business needs of the organization?
Rubinow: Number one, we have migrated our primary production facility from a good building to a really good building here in Florida. That move was recognizing the need for the increased security, increased reliability and all the kinds of things that you want to have in a data center. What I point out to people all the time is, running a good data center isn't the core of what we do. We are not a technology provider in that sense. And running a good data center is not a competitive differentiator. I would say that anybody that's good in this business is going to run a good data center.
We have to make sure we're as good [in data center operations] as we need to be, and on top of that, it's all about flexibility and speed of execution. We've got to be able to change what we do quickly because the time frames in which people measure things are getting smaller and smaller. That's the nature of the digital world. If we want to put something out in production, it needs to be able to be done more quickly than we ever did before. And conversely, if we need to take it apart because it no longer serves the purpose, we need to be able to do that quickly, too. It can't become a major project because we can't afford to spend the time doing such things.
Transform: In many companies, there is tension between the business side and IT because IT can't really respond fast enough to business needs. Have you gone through that stage?
Steve: I've worked in a bunch of companies and a bunch of industries. It's the universal curse of IT. They're never ‘fast enough,’ and they're never ‘cheap enough.’ It’s because people don't understand why it takes so long to do things. In some cases, they're right to ask that question, because it takes longer than it should.
Transform: Speaking of choices, there are a lot of ways to go about collecting data, storing it, processing it, running applications. And it seems like there's no shoe fits every foot. What are you using now?
Rubinow: The migration that I spoke of earlier was from an on-premises site to a co-location site. And it made all the sense in the world because not only were we going to a more robust facility, but why do we need to manage these ourselves? It's a distraction. It has to be done well, but we are not going to do it any better than an excellent provider and especially people who do this for a living, have the scale and the efficiency in the processes and all the trappings of it.
The cloud is a part of what we do. We've done it in bits and pieces here because, in a lot of enterprises, the cloud crept in as an experiment and then it lingers for a while. Much of what we've done with the cloud is what I would call a traditional use – perhaps we want to spin-up something quickly as an experiment for developers primarily. We want to see how it can work and if it works, then we'll make a decision on how to make it more robust. Maybe we just keep it in the cloud, maybe we'll bring it in-house. So we use a combination of both of those, and that is a shifting landscape. I will tell you that in the future, we're probably going to do more in the cloud than we do today. Will it be 100 percent? Probably not, but it's going to shift somewhat.
Transform: Catalina operates across different regions. How does that affect your handling of data?
Rubinow: One of the things that we have to be very conscious of in our business is where the data is travelling from and to – where is it been generated, where is it ending up and what are the laws concerning that? Now, in the US, that’s not so much of an issue. Customers may have individual preferences and they'll express them to us and we'll respond as we need to. But we have a lot of business in Europe. The European Union folks don't like their data to leave Europe, and we have to be conscious of that.
In a pure cloud model where you really don't know where the stuff is because it could be in any of the cloud providers' data centers in the world, it’s hard to comply with international data requirements in terms of crossing geographic boundaries. We have to be conscious of those things whereas perhaps maybe other people with more generic data needs wouldn't have to think about it.
Transform: Let’s turn to disaster recovery. What happens if your co-lo goes down? What's your backup?
Rubinow: In the data center that we moved to, we made sure that it was of a sufficient tier level and we said, ‘They've done the things that we need to have them do and, if there are any single points of failure, we don't know what they are.’
Should something happen, we have a distributed sort of architecture that could be self-sufficient in case the central data center goes down temporarily. Now it can't last forever, but it gives us a little bit of breathing room.
We have some facilities today that we’ll be enhancing in the future. I would say part of the transition in our thinking in terms of more flexibility is in the in-store model, which is our heritage; we had a lot more flexibility in terms of recovering from hiccups in the system. That was because time didn't move as quickly in the in-store model. In the digital world, if there's a hiccup in digital systems, potentially, 3 billion people may notice. Anybody on the Internet could have seen it. That's a big exposure.
Transform: Will those future changes include operations in the cloud?
Rubinow: Yes. We're considering them for disaster recovery reasons that we are probably never going to invoke. The cloud model may not have all the flexibility we want and all the controls we want and all the bells and whistles we want. That’s especially true when it comes to our big data environment which requires more than commodity hardware. Not every cloud provider – in fact, hardly any – will have that.
But for most of what we do, the cloud is a perfectly good secondary model because it's low cost and it's flexible. We don't have to invest in a lot of hardware because somebody else owns the hardware. We just use it as a service. So as a secondary, it's a perfectly good model, and we are in the process of vetting vendors. We have one we use in the United States and one we use in Europe, and we're in the process of looking at that as a global perspective to the best of our ability. That will be not that hard to do. We use the same provider globally, and there are only a few people in the space that can provide global facilities like the ones we need in Europe and Asia and the United States.
Transform: We hear a lot about IT staffing. Are you finding it difficult to recruit folks who have the skills required to move into the future?
Rubinow: This is not a Catalina problem; probably, this is a universal problem. We're not Google and we're not Amazon and we're not Uber and we're not Tesla. So we don't have that kind of appeal for the excited young guys or girls coming out of MIT. They want to work at the cool brand names so they can impress their friends. We are in the middle – and this is one of the reasons I am here – of a transition as an industry. We’re going into the digital space where no one has really cracked the code – not Amazon, not Google, not eBay, not Facebook, not us. So it's exciting for anybody in technology to say, "I can be part of the new frontier."
Having said that, most of our operation is centered here in Florida. There are lots of technology people walking around in Florida. Therefore, we compete with a lot of local companies. I talk to the local universities here all the time. They tell me many of their best people, after they graduate, head for Silicon Valley, they head for Boston, they head for New York. They don't stick around Florida. So it is a bit of a challenge to do that. But we have managed to land some good people, and that's good. We can let people operate remotely so we have an office in Boston, we have an office in Chicago, and we have an office in San Jose not to mention Paris and Tokyo. So we've got some flexibility in our model, and we can accommodate good people who find this part of technology interesting and exciting. It's not easy, but we manage to find people.
Transform: I’m getting that kind of answer from a lot of people right now. Some companies say they’ll be bringing in trusted partners to manage parts of their technology. The idea is to take the load off the headquarters staff and shift it to people who have better resources to recruit the best talent available.
Rubinow: That's right. The Tampa Bay area is growing as a technology hub which is in the right direction. When I came down to Florida, I had great aspirations, having come from New York where you've got a lot of talent, but you also have a lot of competition and the high wages that you mentioned. I took a different tack here. What I found is we recruit locally, we recruit from other companies, we recruit from the local schools and we reach out to other cities where there's more different kind of talent and ask them to come down here.
One of the challenges is that people have to make a conscious decision. Florida is not a tech destination traditionally, so they have to come here for other reasons than ‘I think this will be a cool company to work for.’ That's part of the story that we tell, why it would be a good place to live. For some people it makes sense, and some people say, "No, thank you. I'll stay in New York or San Francisco."
Transform: Would you like to offer any other advice to your peers?
Rubinow: I don't know how many of my colleagues would fall subject to hype, but we talk about cloud all the time and the vendors, of course, have been promoting this idea that ‘the advent of cloud changes everything.’ How many times have I heard a certain 'technology changes everything?’ The cloud is a tool in the toolkit. It's not like, ‘Oh, boy, now all my problems are solved. I just have to find a cloud vendor and send all my stuff there and everything is perfect.’
The cloud is a more complex environment even for those who are adept at running data centers. More complexity can introduce reliability issues and can introduce security issues because there's more moving parts. So I would say anybody that has fallen in love with the vendor hype should have a more realistic attitude about the cloud and say, ‘Yes, it is absolutely a very valuable part of the infrastructure and operating equation.’ But don't give it more faith than it deserves. Look at it with a critical eye just like you would with any other technology.
Point of view storyQ&A: George Rockett, DatacenterDynamics
Srini KoushikChief Technology Officer
Point of view storyThe Traditional Model of IT is Dead - Meet Exponential IT
Zahl LimbuwalaCEO and Co-Founder