Can Google Cloud Platform Win the Cloud Battle Against AWS and Azure?
Remember when the phrase “cloud computing” was dismissed as an ephemeral buzzword by everyone from system administrators to CIOs? Those days are long gone, and the cloud has emerged as a vital part of the IT strategy for many organizations. Granted, the cloud isn’t a panacea, or a silver bullet to solve every IT problem. It still has some legitimate drawbacks, and excessive cloud usage may rightly not be an option for many enterprises. All that said, the cloud is here to stay, and IT professionals that ignore that reality may be putting their careers — and the competitive ability of the organizations they work for — at risk.
One of the mostly hotly contested cloud computing segments is for Infrastructure as a Service (IaaS) offerings. IaaS essentially allows customers to have core IT resources — like compute and storage — delivered as services over the Internet or via a dedicated virtual private network (VPN). These resources are housed in massive data centers, run on virtual machines, and can easily be scaled up (and down) based on a customer’s IT resource needs.
The War over IaaS: Amazon, Google, and Microsoft
Three major players currently dominate the cloud IaaS market: Amazon Web Services (AWS), which is considered the market leader, followed by Microsoft Azure and the Google Cloud Platform. (As a quick aside, VMware vCloud Air is largely considered to be an also-ran in a very distant fourth place.) It’s a competitive situation that has lead to an ongoing war of price cuts between the three competitors, with Google slashing prices on the Google Compute Engine just a few days before this article was posted.
For fans of classic muscle cars, it’s like the Challenger, Mustang, and Camaro rivalry of years past: Three strong rivals makes for a hotly competitive environment that will undoubtedly make all competitors stronger, and savvy IT managers and CIOs can reap the benefit in the form of better services and cheaper prices.
An Introduction to the Google Cloud Platform
Like Microsoft Azure and AWS, the Google Cloud Platform is an umbrella brand name that refers to a collection of cloud-based services. Compute Engine is Google’s answer to Amazon EC2 and Azure Compute, while Google Cloud Storage matches up with Amazon S3 and Microsoft Azure Storage. Google Cloud SQL is roughly comparable to Azure SQL Database and Amazon RDS. You get the idea: All three main cloud providers are competing in a number of different areas, and only these three industry giants seem to have the billions in capital and the vast development resources to compete at this level.
Given Google’s massive cloud resources and experience, it seems to be a natural extension for the search giant to compete. That said, Google’s strength and expertise with managing monster data centers and operating services at cloud-scale are offset by relative weakness among developers (compared to AWS) and a much less compelling integration story with hybrid and on-premise resources than Microsoft Azure offers.
To find out more about how Google sees itself in the Iaas/PaaS market, I recently had the chance to interview Brian Goldfarb (@bgoldy) — the head of product marketing for the Google Cloud Platform — to get more information about what Google has planned for their IaaS offering, and how they intend to compete with Amazon and Microsoft. Goldfarb is no stranger to his competitors, as he most recently was head of Microsoft (then Windows) Azure marketing until May 2012, when he left Microsoft to join Google to head up the marketing efforts for Google Cloud Platform.
What follows is a transcript of my interview with Brian that has been edited for space and clarity.
Jeff James: Maybe we could start with an overview of the Google Cloud Platform, since many Petri IT Knowledgebase readers may not be that familiar with the services it includes.
Brian Goldfarb: I actually think it’s important to go back to the beginning, because a lot of the time…there is this common question. There’s definitely a lot of folks who aren’t 100 percent sure about Google being in the Cloud Platform business, and how we’re thinking about the market, so I think it’s important to cover off on the history.
It kind of all goes back to the beginning of Google. I joke, “We’re a search engine company.” We had to solve a lot of very challenging technology problems in order to do some of the basic things that in 2001 weren’t so basic, like making a copy of the Internet, being able to index it, and crawl it, and do things like that.
We had to invent […some basic cloud technologies…] in order to be capable of sorting that much information. The list of technology innovations required to solve Google’s core problem goes on and on.
Being cloud first, and thinking about cloud infrastructure was part of our DNA from the very beginning. The Cloud Platform aspires to take all of the technological innovation, all of the learnings that we’ve had as a business over many, many years in running the world’s largest technology services, and expose those to developers, and businesses, and IT professionals so they can take advantage.
The thing about the enterprise business specifically is that we began this journey with a foray into collaboration and email, and now we’re extending that out to infrastructure and applications. That’s really what you could think of as the core of the offering.
How do we deliver Infrastructure as a Service technology? How do we deliver Platform as a Service? How do we think about apps and IT, and ultimately big data, so that we can help businesses get the insights that they need to make great decisions?
We look at what does it mean to differentiate and how to think about this market. I talked a lot about tactical innovation. I think there’s a secondary thing that comes along with that too, which is how do you deliver technical innovation in a way that has the right price and performance balance so that cloud investments are significantly more attractive than rolling your own [IT infrastructure]?
The journey that I think cloud is on, broadly, not just Google Cloud Platform but all the public cloud platform providers, is how do we react to and deliver value for the legacy business in the cloud?
Google Cloud Platform and Sustained Use Discounts
BG: We can see a great example in technical innovation, but also in business model innovation. You may not be familiar with this, but in March we introduced a pretty unique model around pricing, called the “sustained-use discount.”
Why this was important is because it’s really changing the way people think about pricing in online, with the cloud. Today, this is the current model. We have to do a bunch of work. Companies we talk to on the regular hire people to help them get the best possible prices from their cloud service providers. We don’t believe in that model.
We think that you should get the best price no matter what, and the cloud, philosophically, should be cheaper than anything that you could do on your own over time. With the sustained-use discount we automatically provide you with the best possible price with no interaction from yourself.
There’s a variety of other examples on the innovation side, around how we’re taking things that make Google better. Like, how do we make the network faster? One of the core tenets of the Internet is faster access means a better experience for users. Faster access to applications is critical.
We [pioneered a virtual] networking with a technology called Andromeda internally, that we’re exposing to the customer. Let’s think about how that relates to the product itself and how we’re thinking about the cloud. I think there’s a couple of things that we know for a fact today. First, I think many years ago when I started working on this stuff, the question was always, “Should I move to the cloud?” I think now it’s “When and how, and how fast can I get there?”
The benefits and the realization of the need have become pretty clear across the vast majority of the businesses and enterprises we’ve talked with.
We look at the things they’re looking at doing, that we see the enterprise is thinking about infrastructure and application requirements. How do they take the existing applications and move those to the cloud? More importantly, how do they build those new applications that are cloud-oriented and take advantage of the benefits of the cloud?
Secondly, we see enterprises thinking about data. There’s this explosion of information that’s happening as a result of multiple different disruptions — the mobile disruption, the Internet of Things, just having internal systems generating more data. Then we need to do something useful with that data. I’ll cover those both as we think about the Google Cloud Platform.
Two Infrastructure Models
BG: On the infrastructure side, there are two models. The first is how do we take the existing applications that people are running, whether that’s legacy Linux applications, legacy Windows applications, and make those run in the cloud? I think today we provide reasonable baseline capabilities in our products that makes that possible. We’re working to release more technology to focus on that.
But what we see happening for those types of workloads is there’s the beginning of a change. Are you familiar with Docker, by chance?
JJ: Yes. We actually just posted a story about Docker not long ago…
BG: Awesome. Docker is the company that’s helping bring a lot of energy around a more generic concept called “containers.” Containers are really a new type of abstraction that’s different than the virtual machine that we’re used to in the current view of the world.
When you think about legacy applications, today it’s about taking VMs and moving them and shifting them into the cloud, and running them.
But what’s happening very quickly is this container model is providing both a more efficient and effective way to do things, as well as a more concrete hybrid solution where I could run applications myself, and then I could run applications in any number of cloud providers, eliminating lock-in, simplifying the deployment models, and keeping control for IT professionals.
At Google, we think the container model is going to be the way that really takes the existing legacy application model, as well as future application models, and brings them to the cloud because of that hybrid approach, because of that control that IT professionals can maintain.
We also believe, and from a Google perspective, we have built our entire infrastructure based on the container model for over the last decade. We run approximately two billion containers every week. In a world where everyone’s running containers, we want to compete on who can run the containers better.
We think that’s going to give enterprises a lot of flexibility, a lot of affordability, and a lot of safety in how they get to the cloud in an effective way.
In addition, there’s also this clouds-first model. Thinking about the cloud and the very beginning of application development, how do you make your apps global from the beginning? Responsive? Flexible?
Thinking about distributed computing, and how do we break free from the arbitrary boundaries that we’ve created for ourselves, Infrastructure as a Service and Platform as a Service? We believe, frankly, that the terminology is meaningless.
Really, what you want is flexible control over what you do, and high productivity that scales automatically. You shouldn’t have to choose between one or the other, and that’s the environment that we’re operating in today. That’s the place that we target, individually investing to really change the gambit, change your impression.
JJ: What advice would you give to IT professionals who may still be skeptical about the cloud? They might say: “Well, the NSA could be attacking this stuff. It’s not in my control. I can’t physically see it and touch it, so I don’t trust it.” What would you tell them?
BG: I think trust is certainly a challenge that plagues the entire IT community. It’s not about Google or Microsoft or others. It is a question that the old-school enterprise perspective is applying to the problem. “I can’t touch and feel my equipment. I don’t feel that I can be safe.”
Unfortunately, the premise is fundamentally flawed, because no business — frankly, of almost any size — has the ability to invest in security capabilities. It’s not their core competency. If you were building widgets, your focus should be building widgets.
Our focus, as an infrastructure provider, is on securing infrastructure and we have over 400 engineers and security professionals that spend every minute of every day trying to do the things the bad guys want to do, before they do it, and making sure the systems are secure.
Very few businesses — we’re talking like zero — have the capability to invest in that kind of security individually.
It goes beyond software. There’s a physical component to this. If you can touch your servers, that means I can probably take your servers if I really wanted to. Physical security means actually thinking about controlling access to your data centers, making sure people who are working on the physical equipment have the right background checks and security clearances. We do all of that, and have all of the certifications to prove it.
The proof’s in the pudding. Heartbleed was a pretty disruptive discovery in the security of the Internet. You know who found Heartbleed? Google found Heartbleed. We found it, we fixed it and helped the community. Not a random enterprise. That is fundamentally why you want to bank on servers you can’t touch.
As far as the data access concerns are concerns, there’s a number of ways that we think about solving that problem. First, we’re doing a ton of work to make sure that all of our data is encrypted and encrypted at rest, and encrypted over the wire, and encrypted all the time.
More importantly, one of the features we’re working on is not only to use our encryption keys, but to let you as the user bring your own keys to the problem, that you have full control over who has the ability to access and decode your data.
There’s dimensions across the entire system that fundamentally make some cloud computing more secure, and I’d say Google Cloud Platform is best in breed as far as our ability to provide security.
Google Cloud Platform Advantages? Price and Performance, Developers, and Big Data
JJ: On that same topic of best of breed, what are some of the capabilities that, from your perspective, Google Cloud Platform has that your competition doesn’t? What are the top two or three features of Google Cloud Platform when compared to Amazon Web Services or Microsoft Azure? What do you believe are the strengths of Google Cloud Platform over those competitors?
BG: I think trying to do apples-to-apples comparisons of cloud provider A versus cloud provider B is a very challenging thing to do.
From where we stand, we’re focusing on trying to deliver value across three areas. The first, as I mentioned earlier, is price and performance. We believe that our technical innovation and our system’s capabilities enable us to provide both the best price, the easiest price to understand, and the performance to go along that — that you get the best value.
The philosophy here is that over time, it will be forever and always cheaper to run in the cloud. We think Google will be able to provide on that front
I think the second area where we’re pushing innovation is around developer productivity in particular. I mentioned earlier breaking down the walls of Infrastructure as a Service and SaaS. How do we think about flexibility and control, versus scale and time to market?
We need to make sure we’re providing both of those things simultaneously. We have a collection of capabilities that make that possible, that’s pretty unique in the industry, in addition to the work we’re doing with containers, which gives us even more flexibility in cloud environments.
I think the third area where we’re highly differentiated as far as categorically is concerned, is in big data, which I haven’t talked about yet. We’re at the end of the day, a log parsing company.
We’re a big data company, and if you look back at the roots of Hadoop and HDFS and many of the databases, the standards that people are using today for doing data analysis, they’re all rooted in the academic computer science that Google released to the world.
In 2005 and with [the Google File System] GFS in 2002, and recently released a paper on Spanner, which is our new super-highest capability, strongly consistent, NoSQL database that fundamentally changes the way you think about it.
That package together, really, is another innovation. Which partner do you think is going to create, over the next 10 years, the most unique and differentiated game-changing innovations, and whether they’ll pass it to their customers? That’s Google.
JJ: On a related point, say I went to a server administrator, someone who really hasn’t used cloud services very much, but he’s considering putting some resources into Google Cloud Platform. What advice would you give, specifically to Windows server administrators? Are there a couple key things they need to be mindful of?
BG: Because this is a new collection of skills, and any administrator, whether that’s Windows or Linux or otherwise, is going to need to get comfortable. It’s a slightly different way of doing things. Like you said before, it’s giving up a little bit of control. You can’t touch and feel it.
The cloud model is just a little bit different. I think identifying some of the lower-risk areas, the obvious one is storage, disaster recovery, backups, those kind of things. Very easy to begin taking data you’re using and moving that to the cloud.
The second one where we’ve seen a lot of success for Windows-based system administrators is around big data. They might take a lot of data, get it into the cloud so they can begin to analyze it, parse it, and begin to look at it in interesting ways.
We were talking for a moment about the big data space in general. Today, if you look at big data, it’s a jargon-y term that doesn’t really mean anything, but more importantly it’s super complicated and it’s really expensive.
One of the things we’re focusing on is how we, A, reduce the cost, and B, make it simpler and more accessible. It starts with a back-end store, a place to put your data. Then it’s “How do I then analyze that data?”
We have BigQuery, which is a SQL-based interactive analysis tool. What’s beautiful about that is you could have terabytes and terabytes of data, and literally get the answer to your questions back in seconds. With alternate technologies, it could take hours or days.
At the same time that we’re innovating in this space, we announced at Google I/O a few months ago [a new service called] Google Dataflow, which is a pretty comprehensive data management service…that provides ETL to analytics to data processing manipulation.
With those two things working together, you really get a very powerful sweep of data tools.
In addition to that, we also provide an enormous amount of support for the open source technologies that are leading the charge in enabling data analytics, whether that’s Hadoop being the dominant technology today running on compute engines and a very fast, high performance, and cost effective way.