A few months ago, Google Cloud he started C4A as a virtual machine (VM) powered by Axion, its first CPU from Arm. Now, as the next step in the project, it starts with C4A Titanium SSDs – its local customized disks aimed at improving storage and performance.

With this move, Google is supporting its C4A reputation and offering VMs that can make cloud services possible. real-time processing. VMs, as the company says, combine ultra-low latency and high-throughput storage with an affordable price, creating an ideal package for applications such as advanced databases, analytics and search engines.

Currently, Google Cloud is making C4A VMs with these Titanium SSDs available in products such as Compute Engine, Google Kubernetes Engine (GKE), Batch and Dataproc. Standard C4A VMs are also available for viewing in Dataflow, with support for Cloud SQL, AlloyDB and other services in the pipeline.

What to expect from Google’s C4A VMs with Titanium SSDs?

Google Cloud’s C4A instances usually come with three storage options: Persistent Disk, Hyperdisk or Local SSD. Persistent Disk is a persistent block storage service where performance is shared between instances of the same type. Hyperdisk, on the other hand, offers a dedicated service, supporting up to 350,000 input / output operations per minute (IOPS) and 5 GB / s through the volume – providing a much better performance than Persistent Disk.

However, for some applications, especially those that require local storage, even Hyperdisk can struggle. This is where local SSDs come in, with the Titanium SSD being the newest addition to the range.

The new C4A instances with Titanium SSDs deliver up to 2.4M random read operations per second, 10.4 GiB/s read throughput, and 35% lower latency compared to previous generation SSDs.

Titanium SSDs, which are directly connected to the compute inside the host server, save storage and network services from the CPU, freeing up resources to improve the security of the application and the performance. This innovation is based on Google’s Titanium system. It manages offloading from the host CPU to the silicon operating system, hardware and software on-host and across the enterprise. data centerconnected to the host CPU using the Titanium Offload Processor.

Current changes

At its core, the new C4A family of Titanium SSDs comes with 72 vCPUs, 576 GB of memory, and 6 TB of local storage. Companies can choose between Standard (4 GB/vCPU) and High-memory (8 GB/vCPU). Connection options, on the other hand, can reach up to 100 Gbps.

All of these can easily support the traffic of real-time data processing applications such as web/application servers, high-end databases, data analytics engine and research. Furthermore, it can handle applications that require memory storage, media processing and transcoding with CPU-based AI/ML.

“C4A…offers up to 65% cost efficiency and up to 60% power efficiency over current x86 processors. Together, the C4A and Titanium SSDs deliver industry-leading performance across a wide range of Arm-compatible applications,” Varun Shah and Nate Baum, senior product managers at Google Cloud, wrote in a joint post.

Early adopters see 40% more

Although C4A VMs with Titanium SSDs are only recently available, some early adopters are already seeing benefits from them. This includes big names like Couchbase and Elastic.

Matt McDonough, SVP of products and partners at Couchbase, highlighted how Capella Columnar, which is running Google Axion C4A events with Titanium SSDs, offers unparalleled cost benefits, ultra-low latency and computational power for computing and performance. Similarly, Elastic’s Uri Cohen said the company saw a 40% increase over previous generations of VMs.

C4A VMs with Titanium SSDs are now available in key regions, including the US, Europe and Asia, with plans to expand. Customers can access them through on-demand, Spot VMs and discounted pricing options.

With significant advances in efficiency, power consumption and scalability, C4A VMs with Titanium SSDs meet the demands of today’s businesses, and set a new benchmark for cloud workloads.