Presented by AMD
Competitive advantage today rests on an enterprise’s ability to deliver exceptional time-to-results for business-critical applications or consumer-facing services, ever faster and ever more reliably.
In the background — or in the data center, to be more precise — this digital transformation and the accompanying explosion in company data has thrust CIOs and IT leaders into an era of relentless scaling. Even as organizations grapple with inflation and economic uncertainty, they are facing calls to provide a high-performance foundational compute infrastructure across the enterprise to develop new delivery models and handle new use cases. These include:
- Streamlining operations and reducing costs while enhancing sustainability (by lowering energy expenses and emissions).
- Enabling permanent remote and hybrid working, often with virtualized desktop infrastructure (VDI).
- Supporting AI, machine learning and database analytics, plus new deployment models — notably containerization and cloud native.
- Mining data effectively to deliver insights that drive revenue growth and increase customer “stickiness.”
- Responding and adapting quickly and flexibly to rapid business changes and evolution to enable ongoing transformation.
New demands on the data center, economic uncertainty, inflation running well above recent levels (principally due to much increased energy costs) — these challenges are outside the control of CIOs and infrastructure decision-makers. And, with CAPEX and OPEX budgets under strain, a tempting option in these circumstances might be to hold fire and postpone investment in data center infrastructure — even if that investment would deliver higher performance within a shrinking power, cost and space envelope.
This can be an especially seductive argument when the data center’s servers are already paid for, which is a common fact given that the average age of these servers is 3–5 years. Surely, the argument goes, it’s better to wait a while, reducing CAPEX and avoiding the effort and cost of upgrades.
The older the infrastructure, the greater the cost
The trouble is those aging servers are not cost-free. The performance of older equipment declines over time, while the time, cost and space needed to keep it running rises. Older servers are more likely to crash, causing unplanned downtime and higher maintenance costs. They are far more vulnerable to sophisticated, targeted attacks. Data centers that push the limits of existing power, cooling and space face increased power costs. Over time they will become more unable to keep pace with the increased, changing business demands. So, eking out a few more years from aging infrastructure may save on CAPEX, but at the expense of rising OPEX. It also risks loss of revenue and competitive advantage.
The fact is, when it comes to data centers, kicking IT infrastructure refreshment down the road is not an option. To serve modern customers, the modern enterprise needs modernized data centers that can support simpler, software-defined environments that improve operations, agility, flexibility and scalability with a lower TCO.
The three pillars of data center modernization
CIOs looking to modernize their data centers need to focus on three main pillars. The first is the requirement to harness all the data in the enterprise to deliver real-time, actionable insights the business thrives on. This calls for systems with the highest bandwidth, lowest latency and fastest throughput. Second is driving savings through infrastructure consolidation. And third is reduced energy consumption and a smaller carbon footprint to meet the sustainability targets that are increasingly a component of corporate stewardship.
The good news for CIOs who still have an eye on their CAPEX and OPEX budgets is that all three objectives are achievable using the latest-generation CPUs. These feature huge improvements in the key performance areas of core density and per-core performance, which in turn deliver record-breaking reductions in the number of servers, power consumed and savings in CAPEX and OPEX.
As an example, take the performance of AMD’s 4th Gen EPYC processors on each of those pillars of data center modernization. Optimized for different workloads and segments — cloud, performance enterprise and mainstream enterprise — 4th Gen EPYC processors significantly outperform competitive processors in key tasks such as server-side Java applications operations/second for commerce (by 2.1 times)1 and supporting 2 times the number of ERP users.2
Meeting the virtualization challenge
A key factor in reducing TCO is virtualization efficiency. Increased virtualization performance is an enabler of infrastructure consolidation, which translates into the ability to deploy hundreds more VMs.
With space- and power-critical challenges, it’s imperative to fit the maximum compute into the smallest footprint, and here 4th Gen EPYC processors deliver quite remarkable performance figures. In a typical deployment scenario of 2,000 VMs, enterprises can replace a rack of 17 Intel Platinum 8490H servers with just 11 AMD EPYC 9654 servers. In hard figures, this adds up to 35% fewer servers, consuming 29% less energy annually, and cutting the enterprise’s annual CAPEX by up to 46%.3
High performance like this not only enables CIOs to reduce both CAPEX and OPEX, it also plays directly into an enterprise’s drive for greater sustainability and lower carbon footprint.
The cost of waiting is increasing all the time
Updating to the newest generation of CPUs can improve a data center’s TCO. The latest CPUs are more efficient, allowing IT leaders to provide the same, or greater, level of performance with fewer servers, resulting in lower costs overall.
As much as IT leaders are concerned about increasing CAPEX costs from servers that are already paid for, and adding to the problem with upgrades, the cost of doing nothing will very soon overtake the cost of modernizing.
That means the cost of waiting is steadily increasing all the time.
Robert Hormuth is CVP, Architecture and Strategy at the AMD Data Center Solutions Group.
Footnotes:
- SP5-104: https://www.amd.com/en/claims/epyc4#SP5-104
- SP5-056A: https://www.amd.com/en/claims/epyc4#SP5-056-A
- SP5TCO-036: https://www.amd.com/en/claims/epyc4#SP5TCO-036
Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. For more information, contact sales@venturebeat.com.
