We’re pleased to announce that 4th Gen Intel® Xeon® Scalable processors on IBM Cloud Virtual Servers for VPC are officially out of beta and available inside our IBM Cloud Dallas, TX, data centers. Customers can now provision one of the industry’s newest microarchitectures inside their own virtual private cloud and gain access to a host of core compute enhancements, including more memory per vCPU than ever available on IBM Cloud.
Are these virtual servers right for me?
Let’s explore. IBM Cloud Virtual Servers for VPC are our newest, developer-friendly platform with rapid provisioning, high network speeds and the most secure, software-defined resources available inside IBM. Your vCPUs would be hosted exclusively on 4th Gen Intel® Xeon® Scalable processors (8474c), which we announced first on IBM Cloud Bare Metal Servers just a few months ago.
Depending on your workload performance requirements, your customization options include three, pre-set, NUMA-aligned profile families (your amount of vCPU instances, RAM, and bandwidth). Consistent NUMA topology embedded into every profile gives you a significantly consistent baseline experience. Each profile supports 32.5 GB of high-performance, local, NVMe-based instance storage for easy access to temporary storage and swap space, DDR-5 memory DIMMs, and PCI Gen 5 for faster interconnects. Instances are booted with Open Virtual Machine Format (OVMF) and run in Unified Extensible Firmware Interface (UEFI) mode for added security.
- Compute profiles are best for intensive CPU demands like high web traffic operations, production batch processing, and front-end web servers. They come with a ratio of 4 GiB of memory for every 1 vCPU of compute.
- Balanced profiles are for a mix of performance and scalability—ideal for more common workloads. They come with a ratio of 5 GiB of memory for every 1 vCPU of compute.
- Memory profiles are best for memory-intensive workloads, like large caching and database applications or in-memory analytics. They come with a ratio of 10 GiB of memory for every 1vCPU of compute.
This release also includes the option to provision your own dedicated host if you need to carve out a single-tenant compute node that’s free from uses outside of your organization. Within that dedicated space, you can still customize performance requirements through the above profile options, and you can also create host groups for specific purposes. Because a dedicated host is a single-tenant space, only users within your account who have the required permissions can create instances on the host.
If you’re looking to try some of Intel’s newest accelerators on your heavier workloads, then you’re in luck because they’re integrated into the processor.
- Intel® Crypto Acceleration is designed to help reduce the penalty of pervasive data encryption and avoid overhead challenges from areas like SSL web servers, 5G and VPN/firewalls. You’ll need the latest Ubuntu or RHEL operating system. The accelerators will intelligently run when your workloads run into a need based on the crypto libraries.
- Intel® Advanced Matrix Extensions (Intel® AMX) is designed to improve the performance of training and inference—ideal for workloads like natural-language processing, recommendation systems and image recognition. You’ll be able to take advantage of Intel® AMX by using extensions from your favorite AI frameworks, such as PyTorch and Tensorflow.
Getting started
See all pricing and provisioning options and save a quote inside the IBM Cloud Virtual Servers for VPC catalog. Or open a chat and get some quick answers.