SAN FRANCISCO -- Google beefed up its public cloud this week with a raft of upgrades, but will they be enough to...
move the platform dramatically closer to its biggest competitors in the eyes of enterprises?
Google Cloud Platform (GCP) has steadily progressed in recent years as a viable public cloud alternative to AWS and Microsoft Azure. The company continues to flesh out that strategy here at the Google Cloud Next user conference, with additional GCP services -- many still in early preview -- around hybrid cloud, security, machine learning, analytics, and edge and serverless computing. Now, Google must prove those investments are enough to win over enterprise users.
IoT and edge hardware stack expands
One of the biggest pushes from Google is in edge devices and IoT, a weak spot for all the major cloud providers, because these workloads require compute power closer than remote cloud data centers can provide or have intermittent internet access. Google rolled out the Edge TPU, a dramatically scaled-back version of its Tensor Processing Unit (TPU) chip that can do machine learning inference at the edge, as well as coordinate with the new Cloud IoT Edge software stack to perform heavy machine learning modeling on the public cloud.
Google's edge and IoT strategy is comprehensive, with a full stack that now supports hardware to run on devices and gateways, and it has enough clout to convince vendors to incorporate its hardware into their devices, said Ezra Gottheil, an analyst with Technology Business Research in Hampton, N.H. But it remains to be seen if that'll be enough to catch Microsoft, which leads the way on edge with its ecosystem.
"The key differentiator for Google is [its] hardware is more comprehensive. But Microsoft has an enormous install base, and that's a lot of leverage," Gottheil said.
BigQuery gets bigger
BigQuery ML, currently in beta, is the latest effort by a major cloud provider to simplify AI and make it accessible to a broader IT audience. Typically, machine learning is done with languages such as R and Python, but BigQuery ML extends some of that functionality for analysts accustomed to SQL. The feature can be used to run predictive analytics models on structured or semistructured data inside BigQuery, rather than load them into a different machine learning service.
Other features added to BigQuery, Google's data warehouse service -- arguably the first to truly set the company apart in the cloud market -- include clustered tables to narrow and speed queries, the integration of geospatial data from Google Earth Engine, and deeper connections to Google Sheets and Google Data Studio.
Security step-up in GCP services
Cloud security is always a foremost concern for companies, and several additions to GCP expand its capabilities there.
Context-aware access adds granular control over access on Google Cloud or third-party applications by defining users' identity, location and request context. These capabilities are limited to VPC Service Controls, but will eventually expand to other GCP services. There's also a Titan Security Key, which relies on firmware developed by Google.
Shielded VMs, currently in beta, monitor and react to changes in VM baselines and runtime state, while distributed denial-of-service protection service Cloud Armor added beta availability of geo-based controls. Other additions include tools to validate container image signatures and scan container registries for vulnerabilities, a cloud-hosted security module, and a FIDO security key with more Google-developed firmware.
Overall, Google's approach to security shows it's embraced the enterprise market, with capabilities that are less about monetization and more about providing a cloud platform to address customers' particular business needs, said Fernando Montenegro, a 451 Research analyst.
"The message Google is coming out with is [that] security is the foundation for everything, and they've kind of sprinkled it all around," he said.
Google's uphill battle to public cloud prominence
The Edge TPUs and their potential to improve IoT experiences through machine learning stand out among the latest GCP services, along with the Kubernetes-led hybrid architecture and the alpha release of the Knative middleware toolkit that converges container architectures and serverless, said Jeffrey Hammond, a Forrester Research analyst.
"There is still work to be done on the 'Why Google' message with regard to other megacloud providers, but the fundamentals are there that would allow them to articulate that message around open source and the convergence of programming models across containers and serverless, as well as core cloud and edge in AI training models," he said.
Tom Petrocelliresearch fellow, Amalgam Insights
That "Why Google?" question looms over everything Google does in cloud. Despite the glut of upgrades, Google executives said they don't strive for feature parity with other cloud providers, but to deliver a set of capabilities that help companies move to the public cloud to solve particular problems, as well -- if not better than the competition.
That strategy requires Google to double down its arguments that its technology is the better answer. Google now generates more than $1 billion a quarter in cloud revenue, though that figure includes G Suite, making it difficult to ascertain the sales of GCP. Still, that's less than a quarter of what AWS generates. The question remains whether Google can ever catch Amazon, or if it will be content as a top option as more enterprises embrace multi-cloud deployments.
"They don't have the early mover benefit AWS has, and they didn't have all those .NET programmers that Azure did," said Tom Petrocelli, research fellow at Amalgam Insights, an IT consultancy in Arlington, Mass. "They're coming in with a pure open source focus, which is fine and I'm a believer in [it], but I'm not seeing anything here that's going to propel them past the other two right now."