Klemsy - Fotolia
Google Cloud has introduced virtual machines optimized for bigger memory and compute-intensive workloads that broaden customers' options for specific workloads and keep pace with AWS and Microsoft.
Like its rivals, Google has long offered compute-optimized VMs geared for applications that demand high throughput, such as high-performance computing, single-threaded legacy applications and gaming.
Customers can now spin up compute-optimized VMs that contain up to 60 virtual CPUs, 240 GB of memory and 3 TB of local storage, Google said. The VMs, powered by second-generation Intel Xeon Scalable processors added to Google's cloud in October 2018, give customers options that match up better with specific types of workloads than general-purpose VMs.
Meanwhile, the memory-optimized VMs focus on workloads such as the SAP HANA in-memory database. HANA's architecture brings transactional and analytic data into main memory, which boosts performance compared to reading it off of disk or even flash storage. Memory-optimized Google Cloud VMs with up to 4 TB of RAM have been available since July 2018, but now customers can tap ones with up to 12 TB of RAM.
The new compute-optimized Google Cloud VMs are now in alpha, priced from $0.209 to $3.13 per hour depending on the size. The memory-optimized VMs will be available this quarter to early access customers, with pricing disclosed at a later date, Google said.
Google's move reflects customer needs, VMs' continued relevance in the cloud
The addition of optimized Google Cloud VMs is not earth-shattering, but nonetheless speaks to customer demand, said Gary Chen, an analyst at IDC. As the cloud expands, customers want to run all kinds of different workloads there.
"There's a fair amount of customers on these clouds that are on the cutting edge," Chen said. "They're chasing every bit of performance they can."
AWS and Microsoft also offer memory- and compute-optimized VMs, and it's hard to pick a winner, Chen said.
"One cloud may be a little ahead of the other in rolling out one type of CPU or another, but eventually they all get there," he added. "The real difference is with the other services they offer."
Gary ChenAnalyst, IDC
Google, for example, has bona fides in machine learning and AI, and these added VM options can cater to those, Chen added.
Modern IT shops embrace containers as an infrastructure model, but VMs have been the bedrock of cloud computing since AWS offered up Xen-based EC2 instances in 2006, and optimized Google Cloud VMs underscore the technology's continued relevance.
"This shows that VMs aren't standing still amid current trends such as DevOps and containers, and VM design is being lightened and optimized as it is influenced by containers," said Jay Lyman, an analyst at 451 Research.
Moreover, while containers have disrupted and replaced VMs to a certain extent, in other cases customers run containers on top of VMs, Lyman said. VMs in various forms likely will coexist alongside containers into the future, he added.