Change is never easy. And up against economic headwinds, even the most worthwhile investments in change can feel hard to justify.
When you're considering a new car or an updated data center, it's tempting to stick with what you have. You may not have the most powerful or efficient option, but it's already paid for and “gets the job done”.
Doing nothing, however, comes with its own set of “costs.” Whether you're driving on a busy street or managing business-critical workloads, you can't afford to forgo important security updates or gains in fuel efficiency or safety. Eventually, your limitations are going to keep you from your ultimate destination or delay your arrival.
Facing an uncertain market and working with tight budgets, it's not surprising that IT administrators feel an urge to hunker down, to rely on their existing systems. Those systems, however, are in need of an upgrade. The average age of a server in the datacenter is 3-5+ years old. Older systems can be more costly to maintain, have a higher security risk, and are often maxed out in available performance. Older systems are less performant and typically have lower core counts - meaning you need more of them. More servers typically consume more power, require more cooling and more space. This is pushing many data centers to the limits of their capacity, in terms of space and/or power and cooling - making it impossible to support new demands of the business.
There is also the perception that since older servers are paid for, that they are “free.” However, the annual cost for power, cooling, and maintenance to keep aging servers in service may be more than the cost to upgrade - and that doesn’t account for “soft costs” associated with older gear like increased unplanned downtime and increased security risks.
And perhaps most critically, older systems simply can't keep up with increasing business demands. The world is changing -- the world of IT has arguably changed more in just the past couple of years than it has in the past 20 -- and now is the time to change with it.
With little competition in the server CPU market, datacenter innovation stagnated through much of the prior decade. There wasn't a pressing need to replace servers; as over the course of four or five years, the performance gains were marginal.
Now, however, IT administrators have a lot to consider. AMD and its OEM and ecosystem partners have brought out a new generation of platforms that are game changers. AMD has led the way with new processors, such as the AMD EPYC 9004 family formerly codenamed “Genoa”, with performance numbers we've never seen before and support for the capabilities needed to make the most of the new CPUs. For example, 4th Gen EPYC CPUs support the newest PCIe® generation (PCIe5), generationally doubling your data transfer rates to help IO devices and storage handle data to and from the CPU. These processors also support the new DDR5 memory standard, with additional capacity and performance to feed vast amounts of data to our high performance, highly scalable CPUs. Containers and cloud-native application programming models are becoming mainstream. Meanwhile, we've witnessed an explosion in new workloads, with AI and database analytics everywhere. These workloads are foundational to business operations and innovation, and they are going to need a whole lot more capacity and performance than legacy systems.
Consider all of these changes together, and the case for a datacenter refresh becomes clear. The question, then, is why switch to AMD?
First and foremost, it's a matter of performance. No matter how you calculate it, AMD offers superior performance -- across a broad range of benchmarks, workloads, and applications. The AMD EPYC family of processors holds more than 300 world records across a broad range of benchmarks, workloads and applications[i]. That includes common business applications, database analytics, infrastructure, high performance computing and more. With aging servers typically offering just 12 cores each, a datacenter using Intel® Xeon® Gold 5118 processors would need around 27 dual socket servers to power 640 VMs (configured with 1 core each). By comparison, you could run those same VM’s with just 5 dual socket 64-core 3rd Gen AMD EPYC powered servers – and get an estimated 24% increase in per core performance, leading-edge security features, and save ~50% on annual operating costs.[ii]
Even as it offers better performance, a new AMD system will also enable you to bring down power bills and aid in meeting sustainability goals. With our advanced “Zen” architecture, we're able to deliver exceptional performance per watt - meaning you could deliver the performance you need using fewer server, and use less power and cooling than competitive offerings.
AMD also offers leading security features that are essential in traditional “bare metal” deployments or in virtualized/cloud deployments. With AMD Infinity Guard, your data is defended, from the moment software is booted, executed, and processes your critical data[iii]. Infinity Guard offers the industry’s first full memory encryption, helping to defend data against certain cold boot and even physical attacks. With secure encrypted virtualization, it also helps protect virtual machines with one of up to 509 unique encryption keys known only to the processor. Most importantly, Infinity Guard doesn't require any code changes to applications to implement.
When changing servers, everything should work easily and out of the box. AMD optimizes its solutions with an extensive list of software and hardware vendors, building deep relationships with partners like Dell, HP, Lenovo, Supermicro and more. Those partners get early access to AMD silicon and products, so that they can test, validate and help ensure that when a customer unboxes them, they have a great user experience.
Meanwhile, AMD has worked with VMware® on a tool so IT managers can seamlessly migrate VMs from Intel to AMD without disrupting business. With the open source VMware Architecture Migration Tool, (https://github.com/vmware-samples/vmware-architecture-migration-tool) the process is quick and automatic.
Those still resistant to change can look to the hyperscalers to understand where the market is headed. The AMD footprint continues to grow in the world's biggest data centers, run by companies like Amazon, Google, Meta, and Microsoft, as well as leading enterprises like DBS Bank, Dropbox, Ford, MasterCard and Wells Fargo.
The bottom line is this: aging infrastructure may actually cost you more than upgrading. AMD delivers the performance, efficiency, security features and simple migration process that makes change easier and more worthwhile than it's ever been before.
______________________________________________________________________________________________________________________________________________________________________________________
[i]AMD EPYC Family of Processors as of 06/19/2023, see amd.com/worldrecords for the full list
[ii]MLNCDN-002: As of 7/20/22 based on AMD internal analysis using AMD EPYC™ SERVER VIRTUALIZATION TCO ESTIMATION TOOL - v10.30 estimating the cost and quantity of 2P AMD EPYC 7763 powered servers with a SPECrate®2017_int_base score of 861, https://www.spec.org/cpu2017/results/res2021q4/cpu2017-20211121-30148.html compared to a 2P Intel Xeon Gold 5118 based servers with a SPECrate®2017_int_base score of 133, https://www.spec.org/cpu2017/results/res2019q3/cpu2017-20190805-16469.html to compare annual costs of existing fully depreciated 4+ year-old Xeon CPU-based infrastructure to acquiring new EPYC CPU-based infrastructure. For details, see https://www.amd.com/en/claims/epyc3x#faq-MLNCDN-002 The Intel hardware is fully depreciated over three years for this analysis. The only Intel infrastructure costs are for operating expenses only: ongoing vSphere support costs, power, cooling, maintenance, real estate. The AMD infrastructure is new, and it includes the hardware purchase that is amortized across the three years of analysis. AMD costs also include annual vSphere support costs, power, cooling, maintenance, real estate. This differs from AMD TCO claims in which the full hardware acquisition cost is taken in the first year. This scenario contains many assumptions and estimates and, while based on AMD internal research and best approximations, should be considered an example for information purposes only, and not used as a basis for decision making over actual testing. SPEC® and SPECrate® are trademarks of Standard Performance Evaluation Corporation. See www.spec.org for more information.
[iii] AMD Infinity Guard features vary by EPYC™ Processor generations. Infinity Guard security features must be enabled by server OEMs and/or Cloud Service Providers to operate. Check with your OEM or provider to confirm support of these features. Learn more about Infinity Guard at https://www.amd.com/en/technologies/infinity-guard . GD-183
