November 14, 2016 | Written by: Peter Rutten
Categorized: Power Systems
Share this post:
At IDC, we’re number crunchers. One number that continues to fascinate me is the server market share from an architecture perspective: x86 versus alternatives. From a vendor revenue perspective, x86 has 81.8 percent of the server market. This dominance is not an ideal situation because the server market has a need for servers with processing capabilities for data intensive workloads. While x86 processors are good, multi-functional processors, suitable for many tasks, they are not specifically designed for today’s heavy data processing.
Servers do a lot of heavy lifting to support the $4 trillion worldwide information technology (IT) market. They are the backbone of digital transformation — supporting digital technologies for business innovation by intelligently leveraging massive amounts of data. The x86 processor has served us well for many years, tracing Moore’s law, but the question is whether it can get us through this new business mandate of digital transformation.
Are x86-based systems distinctly better at ingesting and processing data than competing systems? IDC believes that the short answer is: they’re not.
Multistage IT in the digital era
Digital transformation on a scale-out infrastructure can be looked at as a multistage program for IT.
The first stage, which most businesses today have completed, includes running stateless applications such as virtual desktop infrastructure (VDI) and web applications, creating agile application development environments and moving to a cloud deployment.
The second stage is about opening up the software by adopting Linux. IDC foresees the use of Linux continuing to grow by double digits. Next, bring in open source applications, of which there are thousands, including an open source non-relational database for unstructured data. And thirdly by enabling next-gen application development. By 2018, two-thirds of developers will be developing strategic, next-gen applications. These developers will use different languages, modern frameworks and microservices — almost all open source.
The final stage is about adding cognitive abilities. This allows a system to understand vast amounts of data, and provide recommendations or predictions while learning at the same time. IDC predicts that by 2018, more than 50 percent of developers will be adding data-intensive cognitive components to their applications.
The data-driven nature of digital transformation is creating demand for systems that can take in massive amounts of data quickly and reliably. POWER8 processors, by the nature of their design, do this very well. What’s more, IBM made its POWER8 hardware open source three years ago, which spurred a robust POWER ecosystem to evolve.
The advantages of POWER8
With POWER8, businesses can tap into three benefits:
- These data-hungry Linux-based POWER8 systems can be purchased from multiple vendors, not just from IBM.
- They can be bought with accelerators, GPUs and Nvlinks between the CPU and GPU, which is a little like opening up the sluices on the Hoover Dam to let data in.
- They can be ODMed (manufactured by another party than the logo on the box), so they can be very affordable.
Businesses benefit from diversity in the server market, meaning a healthy mix of multifunctional servers on x86 and servers with other processors, such as POWER8, that run data-intensive workloads well. I wonder how often IT teams evaluate POWER8-based products for data-intensive workloads. Have you?
Read the IDC White Paper, sponsored by IBM, to learn how IBM Power Systems delivers a new breed of scale-out solutions for digital transformation.