20 Minutes With Google Cloud CEO Thomas Kurian At Next '19 UK

‘A lot of enterprises have four or five workloads that dominate their IT landscape, and one of the gaps we had earlier is we didn't have solutions for them,’ says Google Cloud CEO Thomas Kurian. ‘As we shifted into enterprise, we had to provide tools to migrate the existing estates that they had.’


Google Cloud is helping small and medium-sized channel partners through a new value-added distributor program, incentives for landing accounts and by helping protect their territories, CEO Thomas Kurian told CRN.

“In a particular region, we don't have thousands of partners who start competing with one another,” Kurian said during an intimate media round table at Google Cloud’s Next ’19 UK conference in London. “We have been very careful how many partners we admit to the program, so that they get some time to ramp up.”

In addition to partner program changes, Kurian shared his insights on Google Cloud’s enterprise push with new industry-specific solutions, specialized sales teams, economist-led value engineering teams, and pricing tied to successful outcomes. He also addressed European companies’ apprehensions about signing on with Google Cloud, facial recognition technology, remedying the cloud provider’s gaps in mainframe migration solutions and an expanded partnership with NetApp -- whose CEO is his twin brother -- that includes NetApp Cloud Volumes Service for Google Cloud, a fully managed file storage service integrated into Google Cloud to run enterprise and Windows applications.

Sponsored post

Here’s what else Kurian had to say at the ExCeL Center, with the River Thames as its backdrop.

A recurring conversation theme…seems to be this increased push towards enterprise-readiness within Google. You can see that with some of the new hires that have been brought on board from other Silicon Valley enterprise giants. Can you give us some clues to the next...part of that ongoing enterprise push and where your priorities are in terms of bringing enterprises with Google Cloud?

A big shift that we've made this year is solutions. When we talk to a retailer about how can we help their business, it's not about just infrastructure. Many of them say, 'I'd like you to help me figure out how I can improve my supply chain efficiency.' With Vodafone, for example…we're building their customer analytics platform. A lot of it starts with ‘let's try to figure out first what problem you're trying to solve.’ In their case, it was we want to understand what customers are using what product, we want to understand how to upsell and cross-sell them new products. We want to understand what the churn rate is and turnover rate of our customers. That's a business conversation as opposed to a technology conversation. We've shifted a lot of our sales organization to be able to have that conversation with customers. We've specialized our sales teams, we've trained them on how to have that. We have a team of economists that run a team called value engineering, which goes and explains the business value outcome that we're measuring. We even have pricing associated with ensuring the successful outcome, rather than just pricing it by how many cores of compute it's likely to take. That's probably the biggest long-term change we're making to the organization. One example of something we've done to enable that, is we specialized our go-to-market organization around industry. So if you're a CIO (chief information officer) of a bank, you get someone who's only selling to banks and so understands the industry better as well.

What are for you, as a CEO, the hottest topics you have to deal with in the boardrooms with the customers you're talking to? What are their main reasons not to choose either Google or a cloud solution?

One of the biggest topics that have come up over the last few years here in Europe is the concern around, as an American company and as an American cloud provider, can European customers trust their data and systems on a cloud, particularly given some of the regulatory changes. So we’ve have to build a bunch of technology…to isolate data in Europe, secure it with European employees, ensure that even if you get regulatory requests from the government, that the customer has control over whether their data is given up. That is an extraordinarily common question that we get in Europe, particularly from regulated industries like financial services, communications and public sector agencies. So that was a difficult conversation, and it took us time, but we've built an engineering solution for people based on that. It is still a common question we get almost in all major, large European organizations...and we get that as a question about is cloud ready for me, because of the anxiety about moving data into a public cloud.

The Google platform (is) suddenly very popular with developers and people looking for innovation. In enterprise, you guys have been a bit behind, and it's something you're trying to change. Apart from the sales structure and the way you're going to market...could you elaborate a little bit on what the main reasons would be today for companies to pick your competitors over (Google Cloud)? What would you say are your main shortcomings that you would like to work away in the future in regards to the Google Cloud Platform?

A lot of enterprises have four or five workloads that dominate their IT (information technology) landscape, and one of the gaps we had earlier is we didn't have solutions for them. If you're running SAP, we have solutions now. If you're running VMware, we announced an acquisition and a solution for VMware (CloudSimple). If you're running a true legacy environment which cannot even run on a hypervisor, we offer bare metal to enable those. People have asked us do you run Windows properly? We do that now. There are tools that we offer for mainframe migration. A cloud-native company does not have a legacy, and so we used to largely sell to them. As we shifted into enterprise, we had to provide tools to migrate the existing estates that they had.

When you talk to smaller and midsize channel partners -- not the largest systems integrators -- what's the feedback you're getting from them on how you're doing and the restructuring of your sales organization and how they fit in?

We work with many smaller channel partners in SMB (small and medium-sized business) and in corporate customers. We have a very large SMB business. We have more than two to five million SMB customers, not just using G Suite, but also using Google Cloud Platform. Corporate segment for us is from hundred million to up to a billion in revenue. And there, too, because we don't have as many salespeople, we work very actively with partners. We have specialized our partner program for that segment of the market. To reach more smaller resellers and solution providers, we now have a VAD (value-added distributor) program, which is a two-tier distribution model. We did not have that before. We offer incentives now, which are different in the corporate segment and the enterprise segment, for example, to allow channel partners to land new accounts and also to expand an existing account where Google may have sold the initial deal, but we will give it to the partner. That is also new. Many small companies, they are great companies, but they don't have IT people in their organization, so they want a fully managed service, where they don't have to operate even the cloud. And so we have built a new managed service provider program for smaller companies. The feedback we've generally received is that they like the commercial incentives we're offering, and they're seeing fast growth in their business as we scale up the overall Google Cloud business. We've also been very careful to not expand the program too wide. In a particular region, we don't have thousands of partners who start competing with one another. We have been very careful how many partners we admit to the program, so that they get some time to ramp up.

This morning you announced a bare metal offer. Is answering licensing issues from your customers the main reason behind this announcement?

It is to solve three problems. The first is there are some true legacy workloads that if…you have to migrate onto a virtualization stack, it won't run. If you have a very old version of Windows, or you have an extremely old version of Linux, telling them you can move it, but you have to upgrade, will cause real friction, because they have to upgrade the application and they have to re-test it. So the first one is for the true legacy. The second one is we have the largest virtual machines in the world configuration-wise. We support up to 12 terabytes of memory in a single virtual machine. But there are…customers who run SAP HANA, which is their database, in configurations even larger than that --18, 24 terabytes. So to support very, very large environments, you do need a bare metals environment. The third reason is for cases where people are worried about licensing. Some true legacy products do not have a virtualized licensing model. You have to count the number of physical computers, and bare metal lets you count them much more easily.

Your rivals Amazon and Microsoft offer facial recognition tools to third-party businesses. Google Vision limits that to celebrity face-matching. Obviously, Google can do this. Is a move to offer a more powerful service to others imminent and. if not, why not?

So far, we have not offered it, because in some countries, it is not legally allowed. In some parts of the United States, for example, facial recognition is not allowed, because it is seen as a potential source of discrimination. It would be difficult in a cloud environment, where we offer a service that can be used anywhere in the world, for us to be able to do that and remain compliant with different countries' restrictions. If you are, for example, an American company operating, we don't want them to be able to say, 'Well, even though I'm not allowed to use facial recognition in the United States, I just access it in a Google data center in the UK.’ That would be a violation of the legal principle behind facial recognition prohibition in the United States.

You want to be known as the very secure organization, and that's mission-critical to the company. (You're) introducing tools to allow users to actually secure their sites on Google, so that you couldn't access them. It struck me that really ought to be the default...rather than having an option to actually take your encryption keys out.

By default today, if you use Google Cloud Storage, just as an example, you are encrypted by default. You manage the encryption key, we have no access to it. The biggest issue that people find – when…(it’s) is not enabled by default -- is maintaining the key off site from Google. Today we have a solution where you can encrypt the data, you manage the keys, but it's maintained in a repository in a Google data center...to which we do not have standing access. But it is maintained in a Google facility. And the reason we do that is many customers are worried about availability. Availability is if the connection from Google to the off-site location is dropped for whatever reason, because that's going through an internet service provider, we won't be able to look up the key until the connection comes back. What we don't enable by default, is just the off-site, key management. All the all the other functionality -- encryption by default, you manage your keys, no one has standing access -- …is the default position.

What are your main topics and themes in product development for 2020?

A lot of what you'll see is extending our capabilities in the solution space in the industry domain -- retail, healthcare, financial services, manufacturing and media. Extending the core capabilities of our analytics -- the platform, application development, analytics -- as well as broadening our security footprint and then significantly expanding our infrastructure in terms of number of regions around the world.

Is it harder to differentiate yourself? AWS made a whole thing about a Nitro chip and how it enables them to do encryption without any impact on performance whatsoever, and they designed it themselves. I know you are quite able to develop hardware chips yourselves so, in the future, is that something you want to use to differentiate yourself?

We don't try to always develop hardware ourselves, but we do have places where we do our hardware and software. An example is hardware accelerators to perform specific functions. Specific functions could be...encryption, decryption algorithms, cryptographic algorithms...the network interface cards that optimize our software-defined network. Those are things that we have built for many, many years. So when new announcements come up, sometimes they are things we've already done. We build also this specific processor for AI called tensor processing unit, because we see that the number of AI cycles that people are consuming is growing, and we want to be more efficient in offering them cost-effective infrastructure for AI. But we work with all the other providers. We work with Nvidia, Intel, AMD, so we're not trying to compete as a hardware provider. Where we see a specific need that is not met, we'll implement something. But generally, we work with existing hardware providers, and we're not trying to compete with them in any way.

Google announced an expanded NetApp partnership yesterday. Is that something that originated with you and your brother, and do you talk about how you can expand that even further going forward?

We obviously have a good relationship with NetApp that preceded my joining Google. That project, Cloud Volumes, has been underway for over eight months, and it's part of the same theme of bringing enterprise workloads and data to the cloud. I mentioned VMware, Windows, mainframe, Oracle, SAP as all different kinds of workloads. Many companies run NetApp as their infrastructure in their data centers. And when they try to move workloads to the cloud, they say, 'Hey, I really want to run these workloads without having to retest everything.' Because the biggest issue when you migrate workload to the cloud, is risk and time. If you change the infrastructure substantially, you have to retest a lot of things. So with NetApp, we built this ability. You can take your existing storage, extract data from it or copy the data over, you can mount a set of NetApp Cloud Volumes to your compute. And because it exposes the exact same interface as on premise, you do not need to do as much retesting. You can validate and go live. It's a common requirement for customers, and that's why we want to offer it.