Summary
Google has turned a corner. Once full of excitement and perhaps a tad arrogant, the company has tempered itself, reframed the value it offers as an open cloud provider, and refocused its enterprise sales strategy to a partner engagement model. The 25,000 attendees at the Google Next conference in San Francisco showed a new sense of belief in the company.
The 451 Take
Google has turned a corner in recognizing that the enterprise market doesn't just want inexpensive Lego bricks. It demands relationships, integrated products and tooling, and technology that is useable as well as cutting-edge. It expects to win enterprises by marketing an 'open' cloud against other proprietary options. Google still has work to do in building its partnerships and enterprise credibility; scaling to deliver support to all clients will be a challenge. We also think it should revise some of its product tactics. For services where it is differentiated, such as BigQuery or AutoML, it should make sure the market understands that these services are unique to Google, and explain why and how they are better than the competition. For services that are table stakes, Google should be realistic and price appropriately. For example, its IoT Core has less capability than AWS and Microsoft at a much higher price point – this makes no sense. It's better to price competitively on this service, while focusing margin and the value message on those areas where it stands out. The gulf between Google and AWS, in terms of both market and mindshare, is still large, but Google is making significant process.
Context
Google's entrance into the cloud infrastructure was full of promise, but perhaps too little substance. It launched in December 2013, seven years after AWS. But AWS was – and remains – the market leader by a long stretch. Microsoft continues to be a challenger, already embedded with enterprises. Its play is through the integration of software and services. But back in 2013, how Google was going to compete against these players wasn't clear. It had few tangible differentiators at the time, certainly not enough for an enterprise to take a chance with it compared to going with a leader.
Over the years, Google has slowly made inroads and has grown market share. But when it came to differentiators, the message was vague – price performance, containers, analytics were terms it touted, but could rarely justify why it was a leader in any of these areas compared to AWS. Of course, it has experience in scale, container-based infrastructure, and machine learning, but what tangible benefits did this actually bring its customers? It has had its PaaS App Engine since 2008, but enterprise cloud was a new area of development.
This is the year Google turned a corner. Its cloud services announcements and presentations at a packed Google Next 2018 in San Francisco revolved around a number of key themes, which finally give it some differentiators:
- Cloud to ground
- Enterprise engagement
- The value of data
One premier partner we met with described Google as an 'excited teenager that was about to reach adulthood.' Earlier, it had been arrogant and overexcited, and perhaps naive to the enterprise challenge it faced. But tempered by experience, it has raised its game, addressed its failings, and is ready to work. The 25,000 attendees (up from 10,000 last year), many dressed in Google-branded gear, showed faith in the provider is growing.
Cloud to Ground
There was a time when many, including AWS, were saying public clouds would be enough for anyone. But the majority of data points we have collected across all our sources has shown that private clouds aren't going anywhere soon. A mix of clouds, either hybrid (public and private cloud) or multi-cloud, will likely be the default model in years to come.
This has been reflected in Microsoft's strategy with the release of Azure Stack last year, a pay-as-you-go private cloud software based upon, and integrated with, its Azure public cloud. AWS, once a hybrid naysayer, has also claimed hybrid support in the form of its partnership with VMware. The launch of its edge Greengrass software further reflected this softening of hybrid. In fact, the company recently launched EC2 virtual machines on its Snowball Edge, which essentially provides a private cloud capability that integrates with its public cloud EC2.
At Next, Google announced its Cloud Service Platform, an integrated set of cloud services that orchestrate workloads across public and private venues. At the core of the platform is Google Kubernetes Engine (GKE), which has been extended to manage Docker containers on-premises alongside Google's public cloud.
GKE On-Premises is Google's effort to combat Microsoft Azure Stack, but without being prescriptive on the hardware. It is Google-managed Kubernetes software that views an enterprise private datacenter cluster as just another node for management. Policy enforcement is delivered across these venues via the GKE Policy Management service, with monitoring delivered by Stackdriver. GKE Serverless allows serverless functions to be run on containers operating on GKE, and Knative is an open source framework that the GKE Serverless platform is built on. Cloud Build is a managed CI/CD service that manages the build, test and deployment workflows for containerized applications running on GKE.
Last year, Google announced the release of the open source Istio project, a service mesh that enables orchestration of micro-services. Now, Google has announced a managed Istio service that balances traffic across clusters, authenticates endpoints, and integrates with Stackdriver and other Google services. API management for Istio is provided via Google's Apigee acquisition. Additionally, an Application Marketplace lets buyers purchase and provision software to their venues.
With private cloud vendors extending their models to public clouds in a 'ground to cloud' model, Google is offering the ability to embrace hybrid infrastructure via a reverse path of ground to cloud, with a particular focus on open source technologies, specifically Kubernetes and Istio. Azure, too, is embracing this hybrid model, albeit with less focus on open source (although its GitHub acquisition suggests Microsoft's mindset is changing as well).
AWS takes a more conservative approach to hybrid – it does support it via Greengrass, VMware and now EC2 on Snowball, but the company's philosophy appears to be that edge locations are aggregators and processors of data that, ultimately, will be fully exploited in the public cloud. Conversely, Google and Microsoft are viewing hybrid as a longer-term choice for enterprises, with both on- and off-premises locations having specific benefits. A report on infrastructure announcements will follow. There is a trend toward on-premises infrastructure being managed remotely by a third party.
We think Google has identified a strategic differentiator that it is now executing on: open source, hybrid cloud capability that isn't just restricted to moving virtual machines between venues. Google's commitment to, and experience with, the underlying containers gives the company the ability to manage serverless and micro-services across venues, with layers of supporting monitoring and orchestration. With such investment in hybrid clouds, might we expect the company to even support multi-cloud? Considering it already supports monitoring of AWS resources via Stackdriver, we don't think this is out of the question.
Enterprise Engagement
Google is a company used to engaging with consumers, primarily via its search engine. Selling to enterprises is a far more relationship-led, long-term, complex process – one Google is not accustomed to. In the early days of Google Cloud, it had the attitude 'build it and they will come.' For startups such as Spotify, the technology was the big draw, but for big enterprises needing reassurance and solutions, Google was a risky approach. Its announcement of Customer Reliability Engineering promised much, but it wasn't clear what it was and who it was available to.
Google, however, has strongly improved its enterprise credentials. At a panel of customers at Google Next, we asked what, specifically, made them choose Google over its competitors. The answer was that Google had tried more; it had given those customers resources and assistance to develop on Google Cloud; and that Google had made an impact on their businesses and IT strategies. It was the human touch that mattered.
Analysts were given a book of Customer Voices – stories from around 30 enterprises including 20th Century Fox, The New York Times, Bloomberg, Johnson and Johnson, and Etsy that aimed to show Google's cloud credentials. With Amazon having other interests such as retail, media and telecommunications, many of its competitors in these areas would rather not spend their money on AWS, and hence Google has a specific retail opportunity.
Of course, the question is how the company can scale beyond the massive conglomerates. It is aiming for a 100% partner attachment rate for its enterprise clients, which we think is a wise move. In fact, we hear Google sales are being disincentivized to sell directly, to encourage partner uptake.
There are still signs of the consumer-sale mentality – a long thread recently published on Reddit and Medium described stories of data being destroyed within days of credit card expiration, although it's likely few of these companies had engaged Google's enterprise support. AWS succeeded by making its services so easy to consume via APIs, GUIs and documentation, that no assistance was needed – Google faces a scaling challenge if customers of all sizes need a custom and assisted approach.
CEO Diane Greene was brought in to address the enterprise challenge, and is having success in winning CIOs' hearts and minds. A report on the partner opportunity will follow shortly.
The Value of Data
For years, Google has claimed its experience in vast-scale data analytics and ML/AI gave it an advantage in the market. What was missing was a clear case for why it was better. But new capability such as AutoML, which aims to bring ML to the masses, is making a case for Google's advantage here. So too do the additions of a GUI and SQL support for its BigQuery database, which aims to open the analysis of huge datasets to a broader audience.
Its own chip, the TPU, is designed for edge ML processing and is a powerful addition to the hybrid story, as is its Edge IoT software stack. Google's price-performance message, which we were always dubious about, was completely dropped from the conference as it refocused on value, specifically in areas where it excels. This is perhaps a wise idea, considering we found Google to be expensive for both its IoT and Serverless platforms, and don't understand why these products are worth a premium price.
Competition
AWS is the cloud leader by a long way in market share and revenue, with Microsoft a distant second. Google has even acknowledged that more Kubernetes and TensorFlow workloads run on AWS than in its own cloud, even though both were conceived by Google. IBM still has a play, although it needs to get better at demonstrating its overall value proposition.
Oracle is a minnow in the cloud today, but has a messaging differentiator in terms of its Cloud 2.0 architecture, designed to take advantage of technology that the hyperscalers missed out on by being early to market. Looking eastward, Alibaba and Tencent threaten. There are also smaller providers such as Joyent, OVH, CenturyLink, Telefonica and Tencent.
