Last month today: GCP in July

You’ll find a whole selection of posts on the Google Cloud Platform (GCP) blog from July, and a few were especially popular, including supercomputer speed news, new CI/CD options, and a new Kubernetes certification. Check out what was new, whether you’re building cloud from the ground up or putting the finishing touches on a machine learning strategy.

Laying the foundation for your cloud

• Last month saw the announcement that Elastifile, which provides scalable enterprise storage for the cloud, is now a part of Google Cloud. Elastifile will be integrated with Cloud Filestore. This means it’ll be even easier to get enterprise workloads, like those running on SAP, into GCP with native file storage compatibility. It will also simplify managing and scaling compute-intensive workloads, especially useful for companies in media, manufacturing and others where large-scale processing can benefit the business.

• Along with storage, database choices matter for your cloud infrastructure. One big decision for modern business is whether to run databases in containerized, distributed environments like Kubernetes. This post covers what you should consider when making this decision. You’ll get a look at the options available for running databases on GCP, including fully managed and do-it-yourself. There’s also a handy decision tree and tips for success when you are running a database on Kubernetes.

Feeling the need for speed

• Cloud TPU Pods are Google’s supercomputers built just for machine learning. They get faster in each generation of hardware–they’re now in v3–and set three new performance records last month in the MLPerf benchmark. Google Cloud is the only cloud provider that’s outperformed on-premises systems when running the large-scale, industry-standard ML training workloads: Transformer, Single Shot Detector, and ResNet-50.

• The Spinnaker for Google Cloud Platform solution launched last month, letting developers create end-to-end pipelines using Spinnaker on Google Cloud easily and quickly. With the new release, Spinnaker can be installed in GCP in a couple of clicks. The Spinnaker for GCP solution also integrates Spinnaker with other Google Cloud services, allowing enterprises to integrate security and compliance within CI/CD. The solution supports deployment to Google-managed runtimes, other clouds, and on-premises.

School’s (not) out for summer

• There’s a new choice in our GCP course catalog: the Architecting with Google Kubernetes Engine (GKE) Specialization. It uses both lectures and hands-on labs, and in it you’ll learn the ins and outs of using Kubernetes, including how it works, how you can use it as a managed service on GCP, and how to configure a GKE environment.

That’s a wrap for July. Stay tuned for all the latest on GCP.