ra2 studio - Fotolia

Google Cloud Build puts GCP spin on cloud CI/CD

Enterprise DevOps shops have yet another cloud CI/CD choice, as well as additional options for managed Kubernetes, function as a service and service mesh.

Google this week expanded its efforts to lure IT shops onto its cloud platform, as mainstream enterprises evaluate their options for public cloud DevOps support.

Google Cloud Build outperforms cloud CI/CD competitors because it caches source code, images and other dependencies to increase build speed, the company claims. Enterprise IT pros say this addresses a subsegment of the DevOps market that wants high-performing cloud CI/CD services for ephemeral application release projects.

"If you have a project that's going to consume a high number of resources but only for a few minutes, Cloud Build seems optimized for those sorts of workloads," said Harrison Harnisch, a Chicago-based staff engineer at Buffer Inc., a social media management platform with a distributed workforce around the U.S., who has used Google Cloud Platform (GCP) for personal projects.

Buffer has invested heavily in AWS, but Harnisch hasn't ruled out GCP. Google Cloud Build's continuous deployment can be integrated with Jenkins' continuous integration tools, or Cloud Build's CI can link to the Spinnaker CD tool, which originated with Netflix, to make it easier to join with existing pipelines.

We currently use Jenkins, and it works, but it's one of the more brittle parts of the system. This pushes us more in Google's direction.
Harrison Harnischstaff engineer, Buffer Inc.

"We currently use Jenkins, and it works, but it's one of the more brittle parts of the system," Harnisch said. "This pushes us more in Google's direction."

However, Harnisch has also begun to evaluate the Brigade open source CI/CD scripting project spearheaded by developers from Microsoft Azure. He likes the fact that Brigade runs inside a Kubernetes container orchestration cluster, and thus can be deployed to any public cloud.

"Brigade could be useful for DevOps pipelines because it doesn't do deployments outside the cluster -- that's a lot more secure," Harnisch said. "Plus I can use the same pipeline, as long as it's based on Kubernetes constructs, and run it anywhere."

Multi-cloud support is still at early stages of the market but will be important for all DevOps service providers in the cloud in the long run to ride a wave of mainstream enterprise adoption for cloud CI/CD and other DevOps services, analysts said.

"Right now it's still early, and generally customers are still figuring out what they get from each [public cloud] platform," said Stephen Elliot, an analyst at IDC. "Google has built its foundation now, but it has to execute at a high level with sales and marketing over the next year."

Google Cloud Build GitHub integration
Google Cloud Build GitHub integration

Managed Istio stands apart

In addition to Google Cloud Build, the company also unveiled a managed-service version of the Istio service mesh at the GCP Next show this week in San Francisco, and declared the rollout of Istio 1.0 by the end of this month. The Managed Istio service is unique among cloud vendors so far, and follows industry watchers' prediction that complex service mesh technology is best consumed by enterprises through a public cloud provider.

Some enterprise shops that tried to roll out prior versions of Istio on their own turned back after they encountered management complexity and reliability issues.

"To upgrade between Istio versions, we [would have had] to take down every system that was communicating through it," Buffer's Harnisch said.

A more stable 1.0 release and Managed Istio look good for that reason, Harnisch said, but he's already begun to evaluate version 2.0 of Buoyant's service mesh product, Linkerd, as an alternative. And Managed Istio probably won't spur Harnisch's team to migrate established workloads from AWS to GCP.

"But for new projects, it could be interesting," Harnisch said.

Google DevOps portfolio vies for attention against AWS, Azure

Most of Google's updated DevOps services, however, are part of a glut of such offerings on the market.

For cloud CI/CD options going up against Google Cloud Build, AWS offers CodeBuild, CodePipeline and CodeDeploy, while Azure has Visual Studio Team Services. Microsoft has folded GitHub into its cloud DevOps pipeline arsenal as well, though Google partnered with GitHub on Cloud Build, too.

Google also rolled out GKE On-Prem this week, which brings the company's Kubernetes distribution to the hybrid cloud environments favored by enterprises. But AWS has partnerships with Red Hat OpenShift and Chef for on-premises Kubernetes and infrastructure automation, and Microsoft has Azure Stack for the same use. Red Hat OpenShift also has a partnership with Microsoft and can run on GKE.

Google hopes to differentiate its cloud CI/CD and Kubernetes tools through partnerships. GKE On-Prem integrates into VMware vSphere, which potentially gives it a foothold in enterprise data centers where VMware virtualization is ubiquitous. Google and Cisco will join forces this year on the Cisco Container Platform for Kubernetes hybrid cloud infrastructure management.

Google has cachet among DevOps shops because of its technical prowess as the inventor of Kubernetes and a pioneer in cloud-based AI and machine learning services, and these features are enough to sway some enterprise shops as they design cloud-native applications.

Smart Parking Ltd., a car parking optimization software maker based in Australia, hosts legacy apps in AWS, but earlier this year moved its updated Smart Cloud platform to Google to run on a Google Cloud Functions (GCF) serverless deployment. The GCF service went GA this week.

Google's AI and machine learning prowess drove the company's switch to GCP, said John Heard, Smart Parking's CTO, in an interview with SearchITOperations at that time.

"Amazon is a fantastic company at delivering commercialization of commodity items," while Google has focused on how to solve big information problems, Heard said.

"It's about putting your technology into the challenge of how you deal with the whole world at scale in real time," Heard said. "I was convinced we needed to move away from doing the same old thing the same old way."

Dig Deeper on Systems automation and orchestration