Recently I had the pleasure to have an extended and insightful conversation with Jim O’Neill, CIO of Hubspot. We discussed a range of topics related to how his business leverages cloud as they grow, his perspectives on the industry at large, and where he sees the space developing.
Gathering Clouds: Hi Jim, thanks for taking the time to speak with me today. So to get started, what’s your view on the current cloud landscape?
Jim O’Neill: I think we’re in the early innings, so to speak, of the whole movement to the true internet backbone of computing. It’s definitely an exciting time. Amazon has certainly blazed a trail on educating IT staff as well as businesses and consumers in the whole concept of variable computing assets versus fixed computing assets. Personally, I feel like all the recent entrances into the cloud is good for business for everyone as far as choice is concerned, be it manifested through different packaging (public private or hybrid for example) customs builds, or ultimately economics. Not every player will make it, some will be knocked out, but we will benefit from the size, scale, and price of computing continuing to go down at a rate we haven’t seen in the past. Traditionally computing power has always gone down, but computing operations has remained pretty steady and expensive. And I actually see that starting to change. It’s very early on, but all of these different acquisitions are a sign of the massive potential of the market itself.
GC: From your vantage as a business decision maker looking at this industry, what are some of the more promising cloud developments you’ve seen that could improve your business?
JO: I don’t think there are so many things that scale for a business unless they are highly specialized or highly strategic on technologies. At HubSpot, we’ve been on cloud in some form since 2007 when we launched, and it’s really allowed us to focus on product and thenoperations secondarily. But very little consideration went into physical implementation for data centers or physical computing complexities, so that’s just been game changing, and could be for a lot of businesses. The things I get excited about are really, as I see it, the only two changes that have happened in the limelight of cloud computing: compute and storage. The whole networking paradigm, especially with the recent VMware acquisitions – that technology has been more rumored to be available in the last few years, so that’s really come in to the spot light. And someone else had this quote, so I’ll do a creative commons on it: the fact that the network is the last legacy part of the original mainframe environment is a true statement, and is very complex – it’s hardware and topology based, and I think that strides in the network management/routing and addressing space will be the next big thing that will benefit from the cloud, so we can look forward to a much more logical versus physical approach. Geographic distribution becomes less of an issue especially as bandwidth gets greater, so I am extremely excited about networking-as-a-service because it solves some of our pain points hosting in the cloud.
GC: How do you characterize what cloud means for your business and how are you looking at cloud as your business is expanding?
JO: I’m going to look at it a little differently. I look at it in 2 ways, a little bit as a function of our business. There’s cloud consumption that powers our products, which people classify as a marketing-as-a-service SaaS platform, which by the way is some peoples definition for the cloud. We don’t buy servers, and I don’t want to say we rent them by the hour, though we use Amazon, Rackspace and other providers, but we have a mindset of fairly instantaneous capacity – and by fairly I mean like within 30 minutes of a developer being ready to ship a new product we have capacity fully provisioned on the public web. So that’s one way to look at it as servicing our own product.
Another way to look at it is how we run our business. Our entire business runs on the proverbial “cloud,” between vendors to email and collaboration tools to various third party accounting packages. So I’d say we look at it as a service for compute and processing for our product, and cloud-as-a-service as a business function, and that’s how we separate the two. It’s been really great for us. But obviously it’s a time investment to get there. People may not realize that the cloud means IT specialists are managing your infrastructure better than you will. System Admins becomes a much more technical and programmatic role to get the most use and value these products than they have been in the past.
GC: What do you need that the cloud doesn’t provide?
JO: That’s a good question. I’m trying to think of things that the cloud doesn’t provide, and it’s a short list. We don’t have a phone system in the cloud. But even with that there are some metrics and scale questions and concerns coming out as proof for potential solutions. But other than that, everything we use is either SaaS- or cloud-based. There’s a couple of internal tools and vendors that don’t have hosted offerings yet, but that will change over time. There’s very little we don’t do on cloud – it’s really the way we run our internal operations. Someone would have to come to me with a very long list of why you can’t run our business in the cloud for us to even consider pulling it in house.
GC: So what are some short coming s of cloud for your business?
JO: There’s a couple. I mentioned your System Admins are in a higher order with cloud. I don’t mean that in a bad way, running stuff in the cloud means that you leave a lot of internal controls because other people are managing the software you’re consuming, and there’s only so much control you have in that circumstance – controls either through web portals or API, and there’s usually very different tiers of service in what’s available through the web portal and what’s available through the API.
We actually have a development team helping to integrate SaaS providers. Some 3rd parties out there – it’s hard to integrate all these clouds, and I see a ripe industry there waiting to be tapped into. You have different data models, different work flows, different security requirements, etc. But there’s a lot of work to do managing across providers. It becomes a programmatic exercise to manage the SaaS platforms at scale, and if you’re integrating software out of your firewall versus something you use internally, you have less visibility and it’s a more programmatic exercise. That one big thing that’s fundamentally different.
The second big thing is SLA management. Every cloud provider we use has some sort of SLA, be it explicit in terms of service or a contract, or implicit in terms of reputation. You have to get comfortable with those. You have to build your business or build your product on top of those agreements. And it’s a least common denominator game – your SLA will often exceed your vendor’s SLAs and you can’t control that, be it Google Apps, or Salesforce.com, Amazon or Rackspace. The truth is it doesn’t really matter – everyone is going to have downtime. You have to look across the integration – you have to make sure you reduce the failure rates to be in line with the cloud platform with the lowest SLA because that is where your other agreements will settle down to, depending on how tightly coupled they become.
GC: With the rate that HubSpot is growing, is there a point you envision growing out of the cloud, moving away from renting towards full ownership?
JO: I think yes, but with a lot of caveats. There’s always a rental model you can make work. And this is a classic question you are asking which is when can you do it better, cheaper and faster than the provider that you’re using. I don’t know if we’ll ever be in that game, because frankly, at the end of the day, we aren’t data center people. Regardless of whether it’s our product or back office systems, and I say that facetiously; it’s just not in our DNA – we’re product people and we’re marketing people, so I think you can always find a vendor who can provide the level of service you want, whether it’s the private cloud (being the most traditional environment) versus virtualization versus total multi-tenancy. I think any company at scales can negotiate that balance between the provider and consumer. I like to think that we’ll never outgrow some form of cloud consumption with a third party service provider, so it becomes question of who’s staffing those services, as well as a question of fixed versus variable costs – I really believe we can get to the right equation there.
GC: Looking back at the cloud industry at large, what’s missing in the industry for you as a business decision maker?
JO: Integration across clouds is one. It’s emerging though. There’s a lot of vendors promising cloud portability. It’s a risky area. If I’m a business at scale, there’s still at thing called bandwidth that’s has implications no matter how much data you have. So I think it’s more about risk management, or “vendor lock-in.” Say my business is on a closed platform in a closed data center with a closed cloud provider that’s public, you still have limited options depending on how big you get. I feel like there’s a lot of buzz around cloud portability, but I feel like no one has yet come up with a compelling solution to be able to move clouds. There also a growing movement around multi-cloud tenancy, where you can move different parts of your business around to different clouds, datacenter clouds, public and private clouds. It’s still pretty immature, still being more tooling than it is reality at this point.
GC: You guys use Amazon. With Google and HP entering the market in a much more pronounced way, do you see any companies really getting in the ring with Amazon?
JO: I do. Amazon was first to market , and they have a massive land share in both the SMB, as well as enterprise and government space, so they have a pretty broad spectrum . it’s almost “too big to fail” now, since there’s so much there. People need to make decisions based on their comfort around whether Amazon can sustain that level of scale operationally. I don’t mean technically. They are amazing, ground breaking from a technical perspective. I just feel that there will be other players in the Tier 1 space – Google potentially being one. They’re tackling a problem I see at Amazon which is outdated specs, frankly. Amazon keeps coming out with really cool ways to skin the same pricing model of the last 5 years but they haven’t fundamentally changed their pricing model – I don’t know if that’s right or wrong, I’m sure they have people way smarter than me figuring that pricing model out. But in my mind, you’re still sort of paying for 2007-based CPU. Google is smart. Theirs is a C for compute cloud and consumers are paying for high compute capacity at a premium. If you look at the pricing, they’ve been smart in comparing it to Amazon both in terms of pricing and speed. Google, like Amazon, is in so many other businesses that it begs the question: do people at really massive large companies really need services from either of these providers – I don’t know. For real production, I mean, not RD, Q&A, etc. where cloud has traditionally been embraced at the enterprise level.
From a broader competitive perspective on winners in the Tier 2: Rackspace, maybe GoGrid. Datapipe is successfully gaining market position. But economies of scale in ten years are going to be pretty compelling and hard to compete with I think it will be a couple of Tier 1 leaders, a few in tier 2 and other businesses running boutique services, more in virtualization than pure cloud.
GC: Do you see a shift in who the primary decision maker is when it comes to choosing whether to adopt cloud?
JO: It’s totally changing. For us it may be different, since we joined the cloud at the beginning of the cloud movement. In my mind IT shouldn’t be driving those decisions. IT should be supporting them, helping instead to run it. We should be vetting it and helping get it launched. But business people now have access to the data and the capital to make these decisions, and I think they will make the right product feature or platforms decisions and as long as there’s good transparency, which I hope the cloud continues to provide. Whether you’re a vendor or a buyer, make sure that from an uptime, security, SLA business risk standpoint, it’s a fit for your business. So I think there’s a fundamental shift. CMOs are choosing cloud when they are going to be buying software for marketing purposes for example. I think there are more and more sales, operations and customer executives buying from SaaS and cloud vendors. And I think it’s great as long as IT keeps up with it. I don’t think they should be a blocker, they should instead look for ways to enable the shift.
GC: Within an organization, what’s the best way to get buy-in for cloud?
JO: I think the best way to get buy in is to articulate what the business benefits are right off the bat. Everyone will look at ROI and cost, but think about amortization: someone in a business finance department is going to be thinking about the cash outlay upfront over X years, but with cloud you’re paying or committing monthly.
You should just be focused on expressing its value through time-to-market and what the potential disruption during the change over to the system (because a lot of people will focus on switching costs as being prohibitively high). LOB execs shouldn’t be focused on what the switching costs are today, but what will those costs be two, three and four years from now. And this consideration includes increased quality of service. Unless you’re in a very niche, very strategic very secure product space you should be able to leverage the majority of these best practices, and that should be the biggest benefit. Focus on your core business outsource where you can.
By Jake Gardner