Google cloud improved in 2015
Google continued to bolster its services in the extremely competitive -- if not very crowded -- market of global public cloud vendors, yet many of the concerns that lingered in 2014 remained in 2015.
Containers and big data were front and center in many of the upgrades to Google Cloud Platform this year. Google also continued to hammer home its message on lower pricing, made some interesting strategic partnerships and took steps to reverse its reputation for not catering to enterprises.
Despite the new Google cloud features, the service leaves some industry observers wanting more. Often viewed as the third hyper scale public cloud vendor, market dynamics shifted in 2015, as Microsoft Azure pushed ahead as the primary alternative to Amazon Web Services (AWS).
Here are some of the most important improvements to Google Cloud Platform in 2015, and a few requests from industry observers on what they want to see in 2016.
Drilling down on cost with new instances
Better matching of services to customers' needs and optimization of their costs were major parts of Google's cloud plans this year.
Public cloud vendors offer dozens of types of machines, but often those don't match precisely with on-premises configurations. With Google's Custom Machine Types, customers can choose how many CPUs and how much RAM each VM uses.
Pre-emptive VMs are Google's answer to AWS EC2 Spot Instances and provide access to instances at 30% of the standard cost. The use of these VMs, however, is largely limited to tests or stateless workloads without strong uptime requirements, as all instances will be terminated after 24 hours and users can be kicked off at any time.
"Innovative price schemas, such as sustained use discounts and per-minute billing, provide flexibility to meet specific user needs and attract customers with overall savings," said Jillian Freeman, senior analyst with Technology Research.
Not-so-cold storage
Cold storage is an attractive option for cloud consumers tired of archiving data to tape, but the cheap alternative still has its limitations in terms of the cost and time to retrieve that data. That's where the company sees Google Cloud Storage Nearline fitting in, as it offers competitive pricing, with data retrieval in three seconds instead of hours or days.
Big data, big plans
Google made a number of moves around big data in 2015, including the beta release of Cloud Dataproc, the general availability of Cloud Dataflow and the second generation of Cloud SQL, which promises better pricing and performance for managed MySQL databases.
The Google cloud features for large data sets are its real strength, said Sudhir Hasbe, vice president of software engineering at Zulily, an e-commerce site based in Seattle.
Dataflow, a managed service for real-time processing for batch and streaming data, went into general availability in August after a lengthy beta.
Zulily used to run its stream processing through Hadoop in Google Cloud Platform, but made the switch to Dataflow early on.
The beauty of that is it's completely run on its own infrastructure, and it only uses what is required," Hasbe said.
Hasbe is also high on Dataproc, a managed Spark and Hadoop big data service released in beta in September, because it dynamically scales as needed. And he would like to see further investment in the product in 2016.
And in fitting with its price-cutting mantra, Google responded to customers' cost concerns around BigQuery in December with custom quotas to set daily limits project wide or per user.
Hello, OpenStack
OpenStack hasn't made much headway in powering large-scale public clouds, but the open source technology has found a footing with large enterprises and tech vendors -- particularly for private clouds.
So, it was a big step when Google became an official sponsor and the first major public cloud vendor to support the project.
Google's support "changed the game" for OpenStack, TBR's Freeman said. Google is betting heavily on its popular open source container orchestrator, Kubernetes, and this allows the company to integrate it with OpenStack and potentially link with other vendors in the ecosystem.
"While we're not seeing mass adoption of OpenStack yet, portability and interoperability are high on cloud wish lists," Freeman said.
"Google was highly strategic in allowing its technology to better work with the broader hybrid IT landscape," she said.
In another interesting partnership, Google offered some of its big data and storage services to customers of VMware's vCloud Air. Google also hired VMware co-founder Diane Greene to head its cloud platform -- a move seen as a good step to improve Google's standing with enterprises.
Containers
Although Kubernetes can be used in myriad environments, it appears to be a big part of Google's cloud ambitions and gained momentum in 2015.
Some of the best improvements to Google's cloud this year happened around containers, said Dave Bartoletti, principal analyst with Forrester Research.
He cited the smoother pod auto scaling with Kubernetes and improved networking of Container Engine, as well as the improved performance and security for Container Registry.
Wishes for 2016
Despite the progress this year, customers and analysts want to see more features from Google to lure in a broader audience.
"I'm hoping to see more hybrid cloud-enabling features from Google to smooth the transition to cloud from a virtualized data center, for enterprise customers and especially IT operations teams who might not have a relationship with Google," Bartoletti said.
Except for Amazon, long-term success in the cloud will rely on the strength of a vendor's application portfolio, which is why Oracle and SAP can still find success once they figure things out, said Robert Mahowald, program vice president at IDC. Getting those types of packaged applications on its platform will be critical to attracting customers.
"Stuff they're going after now is block storage, object storage and app development," Mahowald said. "It's not really a strong production destination, and that stuff is very intermittent and ad hoc, and not long-term like how packaged software is."
Thanks to Trevor Jones for contributing this article. Trevor is a news writer with TechTarget's data center and virtualization media group. Contact him at tjones@techtarget.com.
Comments