Recently we had the chance to have an extended and at times surprising conversation with Will Ballard, CTO of Gerson Lehrman Group, the world’s premier expert network. Our discussion covered a range of topics including the perspective of a technical leader in a large organization, cloud’s benefits and drawbacks, the role of IT in a forward thinking organization, and how Gerson Lehrman Group approaches cloud enabled development cycles.
The is the second of 2 parts. Check out Part 1 here. Let us know your thoughts on Twitter @CloudGathering.
Will Ballard: CTO at Gerson Lehrman Group
Gathering Clouds: So from that perspective, though, cloud sounds to me as though it’s really enabling you to drive innovation at a pace that heretofore has not been possible in an organization, and I would imagine that that’s a broader trend for a lot of organizations who are moving to the cloud.
Will Ballard: Well, the innovation and its ability, what is it really? It’s about the ability to try a whole bunch of stuff and see if it works. That’s the unique point of this whole issue. It lowers the cost of being wrong so you can try more things until you’re right.
GC: The definition that you’re using for cloud, though, is… is different than the one that’s sort of championed by Ken Ziegler [Logicworks CEO]. He believes that cloud is inherently an outsourced service, that if it’s internal and it has cloud-like qualities, it’s not truly cloud because it doesn’t sort of mimic the economics of an outsourced cloud model. How do you feel about that point of view?
WB: I’m back to my main point of if you have burst or variability, I mean that can be true: you can’t mimic the economics because you can’t really do burst on a fixed asset. Almost nobody has that burst need; they have relatively low scale, an application and a spare, for instance. And then, they set 1,000 servers to work on whatever [application]… almost nobody has that problem. Most people and organizations have 2 servers, 2 database servers and simple costs, even when it runs 24/7. The only identifiable benefit I have of running stuff in-house is that I don’t have to pay for anybody’s profit.
GC: From your perspective as the technical leader [at Gerson Lehrman Group], what would you say are the top challenges to organizations seeking to embrace the cloud?
WB: The most profound one I’ve seen is what I’ll call intransigent IT guys who are typically difficult about it because they’re looking to maintain the status quo of their current employment. That tends to be systems and network guys, more maybe so systems guys, rather than on the development side. So, that’s the biggest barrier I’ve seen: people just don’t want cloud because they don’t want it. Now, the latency issue is a real technical one. The people issue is a real issue as well. But some of the real technical issues I’ve been running into is in an app-by-app or a case-by-case basis, you’ve got to think about whether you are going to have latency trouble. Is it going to run quick enough to get the job done? Because if it isn’t, then it’s not practical. When you start talking about cloud, it’s no longer practical to talk about all of your apps being together in one data center or in one cloud or in one location; you have to think of them as being all over the place because they will be. I don’t think people are ignorant of it, that whole security question – it’s not even a real thing. I think it’s people and the latency issue, and while the former doesn’t stop me, the latency one does because there’s not a lot that I could do about it.
GC: So take me through a hypothetical situation: If you wanted to make that jump to outsource everything, what would the process of migrating everything to the cloud and moving to that outsourcing partner look like?
WB: That kind of fear of migration in my mind is just a data center myth. Be it in my data center of somebody else’s, that’s never really a business conversation at all. I don’t need anybody’s permission in order to move the computers elsewhere; they honestly don’t care and we’ve actually moved the computers a couple of times since I’ve been here. Unless there’s some magical ability that I do not understand, the only way that I can convince myself it would be cheaper to go 100% cloud is probably going to involve me removing a bunch of staff, and that would constitute the only real internal approval I would need. When you’re talking about outsourcing to save money, you’re talking about one thing: internal positions almost always being removed and me having to go through the legal procedures and outplacement, layoffs etc. Moving computers is no big deal. But removing a bunch of staff is. For most organizations of course, that would be a serious hurdle. If you could make a cost model accounting for staff where a company could save money with cloud and they don’t have to deal with lay-offs, you could probably sell the hell out of that! You know, I mean, because no leader likes firing people, but everyone would love to spend less money on computers.
GC: Alright, so continuing this hypothetical situation, let’s say you didn’t have to lay off staff and there was a pretty compelling cost rationale for moving to the cloud, what would be the criteria for the type of outsourcing partner that you would need to be there with you?
WB: Price and programmability; which is to say any vendor or partner would have to have more programmability and functionality than Amazon, or I would just use Amazon. For example, we use Salesforce because it does a bunch more than just Amazon. We use Heroku a little bit because its automation deployment model is an interesting layer above Amazon. But I don’t think I would get any meaningful benefit from evaluating vendors based on, “Oh, they have like 3 smart network guys,” and then they put them in front of me and convince me they’re going to work with my stuff. I just know that that’s not how it works out… that’s sort of like the “Rackspace pitch”. So the vendor API programmability is super important to me. I don’t want to have to talk to a person to do computing. I want to have my developers put API calls in their applications and make them deployable. If I have to pick up a phone call or file a ticket to have an infrastructure change, it’s not good enough because I can do that on my own internally, or I can go to Amazon and not ever talk to anybody and just program it. But I’m on the far side of programming everything. There are plenty of dudes in my position who honestly can’t program and they think of things in terms of labor and people and skills. I tend to think of them more in programming and automation terms. So I’m a rare case, from that standpoint.
Heroku: A platform of choice for Ballard
GC: Earlier we talked about IaaS and PaaS. There’s also obviously SaaS. But what do you see or what do want to see as the next phase?
WB: Wouldn’t it be great if there was something to make quick and easy applications along the lines of what you could do in the 90s with Access or FoxPro? Online hosted, so way higher up than what the current PaaS offerings can do, which are really still sort of programmer approachable. But also not specifically task-focused like SaaS vendors have gone with individual applications. I think that ability of the random, one-off, tiny IT application is missing in the ecosystem right now. For any large enterprise applications, I can essentially get an online SaaS version. Overall, I can get the online, hosted SaaS version, and I can get really good IaaS, but there’s not the quick ‘n’ dirty in the middle. The PaaS stuff that’s available right now is, I think, overly targeted to the start-up programmers who are trying to launch a new thing with Parse or Heroku or what-not…And it’s not well targeted at that. But if they had a toolkit online to build a quick application, not have to think about deploying it – they just put it together – put some fields on the screen, make some reports, mail a link to somebody in a different department and they’re good to go, I think that would be fantastic and I haven’t seen an offering like that yet. They’re really the closest thing to SalesForce.
Amazon Web Services: Defining DIY cloud
GC: What do you feel is missing in the cloud computing industry? Not simply from the specifics of what we’ve talked about with regards to PaaS or the cost model, but what’s missing in the cloud technologies themselves? Or what do you want to see more of that’s only burgeoning just now?
WB: Specifically, I use an application development style that tends to run absolutely everything in memory and you basically just use disk as back‑up, rather than most peoples’ techniques that’s very disk-oriented. There are essentially no vendors that offer truly large memory instances in the cloud; the biggest one I’ve seen out there is like 64 gig and that’s trivial. We run systems with a quarter or half a terabyte here, internally. When you start programming where everything is in memory and you sort of spool stuff off the disk and think of it as a backup technique, you’re programming technique changes a lot and you go a lot faster and you don’t have to performance tune it or fuss around. You end up not writing as much code either and that’s one that’s really hard to do in the cloud right now because everyone seems to be focused on this horizontal scale-out thing rather than the memory scale. I mean, the way I program is not the way most people program so the market is sort of chasing the average consumer.
GC: What’s going to make cloud shift in its broadly accepted understanding towards something that is completely embraced, in the same way that we view a laptop, an iPad or a smart phone as just computing?
WB: I don’t know… I mean my sort of consumer answer is the minute I have an iPhone that gets a reliable 10 or 15 megabit connection…if I have a device in my hand that can get stuff anywhere, anytime of any reasonable size, and I don’t have to wait for downloads anymore, then it ceases to be “the cloud” and it’s just my computer. So, when it’s fast enough, you can’t tell the location… and we’re a ways out from that. In a business setting, I don’t even really think about cloud or not cloud like that anymore; it’s just not an interesting conversation. But I know a lot of people aren’t all the way there yet. I think that, in the IT business setting, you’re just going to need a couple of extremely conservative players having done it, to testify to the other conservative players and then it just becomes normal. The other option is that the cost model is so shocking and compelling that your CFO knows about it and comes to you and says, “No, really… you’re doing this; I don’t care if you like it or not” So sell it from the finance side…
GC: Is that [trend] happening in places?
WB: No, I haven’t seen it yet. To be honest, the costs don’t really work out. If you have an IT department that’s hard to manage or your guys aren’t that great, and if you have that particular issue, outsourcing is a nice way to put a bow around it and make it go away. But I’ve never seen it be cheaper, in a full-on managed services scenario. Rackspace, for example, was radically more expensive than what we were spending internally at Demand Media.
GC: Is that just their [Rackspace] pricing model though? There are plenty of competitors who are going to give a more competitive pricing.
WB: There might be… I mean, that’s fairly dated; I mean, it was 2007 and you’re paying for the convenience of being in other markets. In addition, paying for managed services is always real expensive. Practically, they have IT guys too and they might be smarter and better than your own IT guys. With your internal team, however, you don’t have to pay the mark‑up for someone else’s profit.
GC: Take us 3 years into the future in terms of where you feel the cloud industry is going to be. What changes do you see having occurred? Where are the major industry players [like Amazon, Rackspace and others] going to be? Who is going to be coming up in a much more pronounced way against those major players?
WB: My expectation is that 3 years out, the major computer vendors and services companies will be IBM, Dell, Microsoft, HP, who will have almost completely consolidated the small and mid-market players away. Amazon will still be in there. The server vendor type I think understands that that market is potentially not going to be the same shape going forward. So, you have everything from Microsoft Azure to what Salesforce has for its platform. You have Amazon, which is sort of a mix of a raw IaaS and also some higher level application programmability. Then there is Heroku, which is sort of in between. With something like Rackspace, I don’t have any opinion on what it does that’s better to even warrant me to look at it. So, I don’t expect someone like Rackspace that is sort of, at least in my mind, focused on people and service, to be a really awesome thing to add to Dell, who has been steadily expanding in their own managed services and their outsourcing and consulting buys.
I’m sure that the investors don’t go into this stuff for kicks. There’s clear exits painted in every direction for these large, classic IT players to aggregate.
By Jake Gardner