Google has expanded the configurations for its public cloud virtual machines by opening additional use of the platform...
for new and old workloads.
Google this week made Intel's Skylake Xeon processors generally available on Google Compute Engine -- the first public cloud to make the next-generation chip available to customers. The move promises better performance and efficiency and is accompanied by other changes to the infrastructure-as-a-service offering that diversify the VM options for IT shops.
The Skylake processor, first released for testing on Google Compute Engine in February, provides up to 64 cores and 455 gigabytes of RAM. It's now available across all Google cloud VMs but targets applications that require high performance or lots of memory. Its availability is currently limited to the Western U.S., Western Europe and Eastern Asia Pacific regions, though it's expected to be extended to other regions soon.
Lytics, a personalized marketing and customer data platform in Portland, Ore., plans to move its container-based architecture to the Skylake 64 vCPU servers, to house its containers in a single, larger space rather than segmented across multiple machines.
"Just by having bigger servers it allows us to be more efficient and get better utilization rates," said Aaron Raddon, Lytics CTO and co-founder. "You want the biggest machine possible so you can really have as little wasted space as possible."
Lytics also signed up for Committed Use Discounts, which were made available earlier this year and provide significant savings if customers purchase a certain capacity each month. Between that and the move to the Skylake VMs, Raddon expects the company will double its volume at roughly the same costs.
More flexible options on Google Compute Engine
With the addition to the Skylake processors, customers can also select processor types for their Google cloud VMs, potentially a cheaper option for workloads with less intensive demands. This includes the Broadwell CPU, which is now available with up to 64 vCPUs in all regions. Google said it will offer Skylake VMs at no additional cost for a 60-day promotional window, after which it will tack on a premium of up to 10%, depending on machine configurations.
The major cloud providers continue to churn out more instance types to meet growing demand for their platforms, but a feature unique to Google is its Custom Machine Types, which provides custom VMs with vCPUs and memory rations within certain limitations. Google has reduced the memory ratio restrictions with a new maximum 455 gigabytes of memory per VM, compared with the previous cap of 6.5 gigabytes.
The ability to match instances to computing tasks should ultimately help fine-tune deployments and save money, said Dave Tucker, senior vice president of product development at Workiva, a financial services software provider in Ames, Iowa.
"As we build out an increasingly robust set of microservices, we are naturally finding that some services are more specialized and compute-intensive than others," Tucker said.
The increased memory flexibility opens Google Compute Engine to more in-memory, relational and nonrelational databases -- a market cloud providers have aggressively pushed into lately. It's also an area where Google still must convince enterprises it's a good place to migrate more traditional workloads. For example, Google earlier this year added support for SAP -- seen as a positive step, but almost two years after Microsoft and AWS did the same.
Google has been a leader in adding granularity to its platform, as evidenced by its per-minute pricing, said Charlie Li, chief cloud officer at Capgemini, a global business management consulting firm with U.S. headquartered in N.Y. And while the level of customization in the cloud doesn't equal that for on premises, the market will continue to go this direction.
Microsoft made significant gains on AWS a year ago, and now Google is doing the same due in large part to its technical prowess and ability to build everything from the ground up, Li said.
"AWS today leads in terms of the number and types of services, but Google, in terms of the technology and flexibility and pricing; [they] are challenging the others and you'll see AWS and Microsoft respond," he said.
Trevor Jones is a news writer with SearchCloudComputing and SearchAWS. Contact him at firstname.lastname@example.org.
Google launches new IoT service to compete with Azure, AWS
Build a virtual private cloud with Google XPN
Learn which workloads benefit from cloud GPU instances