Premise. Google has been the sleeping giant of the enterprise cloud market. Users and competitors wondered if and when it would wake up. At Google Next 2018, we saw the first impressive signs that the giant is stirring.
Put 25,000 attendees and 3,000 Google engineers in one place for three days and a lot of interesting things will be said. One of the most interesting things we heard at Google Next ’18 was a throwaway comment by Google Cloud CEO Diane Greene at the industry analyst session. To paraphrase:
“We all are very early in this cloud business. No one really knows how to do this.”
Wikibon agrees: We are early; customer requirements are rapidly evolving; business models are immature; there’s a lot of learning yet to do.
However, we also believe that some cloud companies seem to know more about providing cloud services than others. And given that Google Cloud is at best a distant third in the public cloud market, a big question that Google Cloud needed to answer at Google Next ’18 was: Could they catch up?
Short answer? We think so. Slightly longer answer? With the right investment and right focus, Google Cloud has the talent, technology, and scale to be an enterprise cloud force. Momentum is building. In the last 3 years, $30.9B has been invested in Google Cloud. The Google Cloud business unit is adding headcount faster than any other Alphabet unit. The company has expanded to 17 regions, fewer than Azure but comparable to AWS. It claims to feature the world’s largest private network, with over 100 points of presence (which doubled in the last two years) and extensive subsea cabling (which also almost doubled). Over 4 million firms are paying for G Suite (including big companies like Airbus, Whirlpool, and Colgate), which, when combined with free versions, translates into 1.4 billion users. Finally, in 2017, it tripled the number of $1 million+ deals.
To accelerate that momentum, Wikibon identified five domains where Google Cloud needed to demonstrate progress. They were:
- Upgrade the Google Cloud enterprise engagement model.
- Deliver AI superiorities as superior AI services.
- Attract enterprise developers to show a Google Cloud preference.
- Catalyze partnerships that can extend Google Cloud’s capacity to specialize.
- Advance trust in Google Cloud.
Here’s how Google Cloud did relative to each of these domains.
Upgrade the Google Cloud Enterprise Engagement Model
Users haven’t been doubting Google Cloud’s technology. Nor, obviously, its scale. Instead, the two questions consistently asked are: (1) Will Google Cloud be a force in establishing the state-of-the-art in enterprise cloud services? And (2) Will they engage the enterprise like a true enterprise technology player?
Users will be happy with both of the answers that emerged from Google Next 2018. On the question of cloud innovation, this was by far the simplest and most cohesive presentation of the Google Cloud business that Wikibon has heard. In the past, Google Cloud seemed an inside/out packaging enterprise: Here’s a bunch of good cloud technology, let’s package it up and charge for it. At this show, however, Google Cloud displayed an outside/in focus that we haven’t seen before. Google Cloud now is organized around five core businesses (GCP, G Suite, Google Maps, Chrome Enterprise, Android Enterprise), each of which lines up with real enterprise issues. More importantly, the company clearly articulated its value promise:
- More secure by design.
- Most open cloud.
- Leader in AI and ML.
- Cloud-native collaboration that’s easy to use.
All good, but the real user concern we’ve been hearing about Google Cloud is their engagement model: Would GCP craft an engagement identity that users – and partners – love?
Here, too, Google Next 2018 showed promise – real promise. Wikibon believes Google Cloud is establishing a differentiated approach to enterprise engagement that will forge a unique, highly valuable identity. Our observations are that Google Cloud is:
- Parlaying its unparalleled open source collaboration into enterprise collaboration.
- Leading large enterprise engagement with engineers.
- Committing to “hand-to-hand” combat for enterprise business.
- Investing heavily in a value-adding ecosystem.
Deliver AI Superiorities As Superior AI Services
From pure research to production workloads, Google is a universally acknowledged leader in AI technology. Their voice recognition (Google Assistant) and translation/transcription are the best we’ve seen, for example. And, of course, Waymo self-driving vehicles have racked up double the test miles of any other self-driving car outfit. As a company, Google promises to “advance AI for everyone,” but there has been some doubt about how it would do so for enterprise customers.
Google Next ’18 answered this question, as well. Google introduced or extended a number of leading AI services. Among the services we found most interesting:
- Cloud AutoML. Think “automated machine learning,” not something from Waymo. AutoML promises to simplify the process of creating machine learning models for multiple use cases, including translation and language processing, but appears especially strong in the vision domain. Originally discussed in a November 2017 research paper, AutoML uses machine learning to create machine learning models. A simple GUI allows a developer to securely train, evaluate, refine, and deploy models based on private data – that stays private after the process is complete. A number of start-ups (e.g., ParallelM) and hyperscalers (AWS and Azure) are likely to introduce similar capabilities, but Cloud AutoML is the first enterprise-grade service of its kind that we’re aware of. It’s off to a strong commercial start: Google claims that nearly 20,000 customers have declared an interest in Cloud AutoML.
- Kubeflow v0.2. To simplify the task of adding ML capabilities to any Kubernetes-based application, Google introduced Kubflow V0.2, which brings scalable on-demand, repeatable, and microservice-based architectures to ML workloads. Wherever Kubernetes is running, Kubeflow could follow. The technology is part of Google Cloud’s overall approach to hybrid, multi-cloud.
- TensorFlow + hardware. Infrastructure for AI workloads will evolve rapidly over the next few years, but it will settle into an architectural pattern that centralizes AI training (where the models are built) and distributes AI inferencing (where the models run). Tensorflow is a broadly adopted, Google generated, open source project that uses a dataflow approach to process arrays using a graph-based data model. While it runs on multiple CPU architectures, Google has built custom CPUs – called Tensor Processing Units (TPUs) – to accelerate the performance of Tensorflow workloads inside Google. TPU has been available. To improve model training performance, Google Cloud introduced ML services based on Cloud TPU Pods and TPU V3. Edge TPU ASICs are tiny (four can fit on a penny) machines that partners can employ in edge devices. The combination of TPU services, specialized training hardware, and specialized edge-based inferencing hardware gives Google the best end-to-end ML technology story in the market.
- Strengthening cloud analytics services. Google Cloud’s BigQuery is a highly regarded set of big data lifecycle tools that got better, simpler, easier to manage, and better integrated with market leaders. First, a new integration with Google Sheets and a beta version of Data Studio Explorer simplifies BigQuery data exploration and visualization for non-IT users. Second, Cloud Dataproc (in Alpha) streamlines many big data-related tasks, which both speeds time-to-value and can dramatically reduce cloud costs devoted to administrative tasks. Finally, Google Cloud announced that Hortonworks DataFlow and Data Platform are enhanced on GCP by better integrating Google Cloud Storage to improve hybrid cloud deployments and streaming applications.
Attract Enterprise Developers to Show a Google Cloud Preference
Google path to catching up in the enterprise cloud market runs straight through the land of enterprise developers. And increasingly, the enterprise developer landscape is shaped by open source, APIs, and security.
In response, the cloud industry is shifting focus. Infrastructure-as-a-service (IaaS) offerings are starting to take on Platform-as-a-service (PaaS) characteristics as enterprise developers gain market power and shift their focus from simplifying infrastructure deployment to exploiting advanced application services, like AI.
Google is well positioned to catch this wave. It’s a fixture in open source (one of the two biggest contributors, with Microsoft) and features 20,000 projects on GitHub. Its Qwiklabs is a rapidly expanding destination for online developer training. And Google’s incredible market scale ensures that its contributions to the development community have a strong chance to become de facto standard – or even conventions – in the developer world. For example, Kubernetes has emerged as the leading orchestration platform; it’s launching 4 billion containers per week. Spanner is emerging as a great, stable technology for scalable relational database applications (within practical engineering limits of distributed data, of course).
At Google Next 2018, Google Cloud discussed a number of important advances that will excite enterprise developers, including:
- Istio and Apigee API Management for Istio. Istio is an intriguing service platform that is bringing together DevOps, microservices, and multi-cloud. Apigee is integrating Istio into its microservice management platform. It’s a match made for the cloud and brings world-class API management to the Istio platform.
- GKE On-Prem and Knative. Kubernetes Engine (GKE) dramatically streamlines the process of deploying Kubernetes clusters. Google Cloud is making it available on-premises for enterprises that want to run local Kubernetes clusters. Knative complements Kubernetes by providing frameworks that simplify building, deploying, and managing Kubernetes container-based applictions. One especially important pattern? Serverless. Together, GKE On-Prem and Knative are an important step for Google Cloud in the direction of True Private Cloud support.
- Improved tooling for DevOps practitioners. Cloud Build provides a range of automation for application development lifecycles, including code, build, test, and deploy. It also enhances artifact management by automatically scanning builds for security vulnerabilities. Stackdriver Service Monitoring can speed time to root cause analysis and resolution by making Istio and other Google services observable, improving a shop’s ability to monitor service level obligations.
Catalyze Partnerships that Can Extend Google Cloud’s Capacity to Specialize
One cloud business domain where best practices are a work in progress is partnership. Business conventions and relationships for technology and go-to-market partnerships have been well established for decades. Cloud business models are upending many of those conventions and recasting channel relationships.
As a company, Google has been at the vanguard of forging new types of cloud-based business partnerships. Its marketing cloud in particular supports an enormous volume of complex business transactions through a self-service sales and support interface – as well as an expansive array of marketing service companies that act as partners for Google’s marketing cloud services. Google knows how to present complex cloud services to a range of customer audiences.
Will it be able to translate this knowledge into an advantageous ecosystem? Early results are encouraging. Partners want to see commitment to and consistent application of partner programs that pay. Google Cloud is committing to partner participation in 100% of deals (it’s just over 80% today). That includes sales deals, professional services engagements, and marketing programs. They’ve doubled partner-dedicated resources while at the same time reducing the number of partner programs to one (by 2019). Using their leadership open source presence and outstanding commitment to cloud-related knowledge transfer, they hope to build the most collaborative and innovative partner presence in the cloud market.
Two areas of special significance. First, Google Cloud introduced Cloud IoT Edge in Alpha. Using this, partners can embed Google Cloud’s ML services into edge devices that run Android or Linux. It will locally store, process, and capture data essential to running and managing these devices. Cloud IoT Edge will run TensorFlow Lite models locally, and will get a special boost from devices that employ Google Cloud’s Edge TPUs. Second, Google Cloud announced a program that will allow top-tier MSP and other partners to help enterprises run Oracle workloads in the cloud with less than 1ms connectivity cost.
Advance Trust in Google Cloud.
Google is an acknowledge leader in cloud security; Google Next 2018 showed how it will start leveraging that leadership in Google Cloud. The company showcased a number of technologies – some leadership, some catch-up – that can strengthen its security conversations and services to large enterprises, including context-aware access, shielded VMs, and vulnerability scanning. However, the Google Cloud’s real security story is it’s up-and-down, as well as end-to-end security commitments. GCP features multiple levels of security, from hardware up to service delivery. They refer to it as “defense in depth at scale.” In action, it’s impressive: For example, 94,000 web pages scanned for harm every minute and 400 million android devices health-checked daily. Google also trumpeted the fact that there are no reported or confirmed cases of account hijacking on a Google platform.
How will Google Cloud sustain its transactional business model as the rest of the company employs an ad-based business model? Diane Greene, Google Cloud’s CEO, noted the parallel between Google’s overall mission and Google Cloud’s mission: Google’s mission is to organize the world’s information and make it universally accessible and useful; Google Cloud’s mission is to organize information within customer enterprises, keep it secure, and present it back in a valuable way. It’s an ambitious vision, but it’s doing.
