Today, I had an insightful discussion with Tom Krazit at the Gigaom Structure’14 conference, where I shared Intel’s perspective on the evolution of the data center in driving the growth of the digital services economy. We’re currently experiencing a striking industry transformation as IT shifts from supplementing business operations to becoming the nucleus of business activities. This transformation, coupled with the ascent of cloud computing, has raised several pertinent questions about the fundamental principles of data center architecture. Two major shifts we’re witnessing involve the progression towards software-defined infrastructure (SDI) and the transition to scale-out, distributed applications. The acceleration of application development and the swift deployment of new services demand that the infrastructure also evolves. This evolution entails shifting from static configurations to dynamic ones, manual operations to full automation, and fixed functionalities to open standards.
At Intel, we continue our tradition of embracing these transitions, spearheading technology advancements to redesign contemporary data centers to address the future’s needs.
We commenced this journey with our commitment to deliver excellent technology for all data center workloads, covering servers, networks, and storage. Starting with our conventional Xeon processors, we supplemented them with workload-optimized products, like our Atom SoC processors for light web-hosting, Xeon Phi for intense parallel processing, and the Xeon D SoC series for hyperscale environments.
Subsequent, we extended our product assortment beyond these workload-optimized solutions, releasing 15 personalized products in the previous year to accommodate specific client requirements, including giants like Ebay and Facebook. Our development pipeline for custom solutions keeps expanding, doubling the number of products planned compared to 2014.
But what excites us more is our forthcoming innovation in processor design that can significantly elevate application performance through fully custom accelerators. By integrating our outstanding Xeon processor with a coherent FPGA in a unique package, we’re providing a socket compatible solution to our standard Xeon E5 processor offerings.
This announcement is anticipated because it offers our customers a programmable, high-performance coherent acceleration capability to bolster their crucial algorithms. Furthermore, with down-the-wire programmability, the algorithms can be updated as fresh workloads emerge, and computing needs change. Based on industry benchmarks, FPGA-based accelerators can deliver over 10X performance improvements. By combining the FPGA with the Xeon processor, we project that customers will witness an additional 2X performance boost due to the low latency, coherent interface.
Our new Xeon+FPGA solution is another custom choice, and one more tool for customers to enhance their crucial data center performance/TCO metric. It underscores our dedication to delivering best-in-class solutions across all data center workloads and our enthusiasm to lead the industry’s evolution to cloud services.
Beyond silicon, our innovation extends as we collaborate with industry peers to expedite the migration to SDI. Our initial collaboration with Facebook, as part of OCP, to define a rack-scale architecture (RSA) is gathering pace amongst cloud service providers, telco service providers, and hosts. Notably, an emerging trend seems to focus on optimizing solutions at the rack-level, transitioning from singular nodes to multi-node configurations. This shift towards pooled, multi-node solutions, driven by software, forecasts the next evolution of data center efficiency. We are at the forefront of this transformation, leading its implementation.
The early adopters of SDI are network operators. With the constant climb in network capacity requirements and the need to swiftly introduce and monetize new services, the transition to network function virtualization (NFV) becomes lucrative. Intel, as an ETSI NFV Forum member, contributed significantly to the creation of the NFV spec and the development of nine use cases. Presently, 20 PoCs are in progress with carriers worldwide.
Despite the fact that it historically takes service providers years to deploy new services from concept initiation, virtualization aims to reduce service deployment time from years to months, and eventually minutes. For example, Telefonica has plansto turn 30 percent of its network to NFV by 2016. Leveraging over a decade of technology innovation in server virtualization and strong technology ecosystems, we lead the network transformation to SDI.
Lastly, our significant investments are targeted at extracting valuable information from the enormous quantities of structured and unstructured data. We are actively partnering with industry leaders to expedite the delivery of cutting-edge analytics solutions. Our recent agreement with Cloudera exemplifies our commitment to enabling enterprises to unlock their data’s economic value through big data insights.
Throughout our history, we’ve seen how transformation breeds both opportunities and challenges. The transition to cloud services and SDI presents a colossal opportunity, and we intend to seize it. We spend billions of dollars annually on data center R&D to meet our customers’ evolving needs. As we strive to bring our vision of a data services economy to life, we urge others in the industry to join us on this exciting journey.
source: Updated by someone/Intel, 2025
Updated in 2025 to align with recent developments.