Part 2 of a 2-part series. Don’t miss Part 1.
Machine-learning, artificial intelligence, and analytics capabilities integrated with enterprise and mobile applications are set to bring more innovation, changing how enterprises will serve their customers. The following three trends have come together, to make it possible for enterprises of all sizes to apply analytic techniques to business processes, changing how they will serve their customers:
In addition, Google recently released a machine-learning service for developers. While IBM released three new Watson APIs, allowing developers access to technology for sensing emotional and visual cues, delivering insight on tone, emotional context, or sentiments of voice, text, or images. Microsoft launched a new machine-learning PaaS service, supporting Hadoop and Spark, with languages Python and R, providing developers a high-performance, real-time application-execution platform with pre-built components designed for predictive analytics solutions.
The resulting mobile apps and on-line websites consolidate decision making by people who need to act in real time, where insight from nontraditional data sources is infused in business processes. Several examples include the following:
According to the latest information from IHS, 22 percent of physical servers, 17 percent of virtual machines, and 12 percent of Linux containers, on average, are expected to be used in off-premises data centers by 2018. When this happens, there will be opportunity for hardware and software innovation that drives the use of bare-metal servers and switches, combined with open-source software.
Microsoft Azure has already indicated 90 percent of its new servers are running open source software. Goldman Sachs has stated that it will continue to migrate certain applications to the cloud, while increasing its purchase of bare-metal servers running open-source software for its data centers. Goldman Sachs has stated that more than 80 percent of the servers it has acquired since last summer are based on Open Compute Project (OCP) standards, and its OCP server count has increased from 2,500 to 4,300 since last year. This number still represents a small fraction of the 125,000 servers in Goldman Sachs data centers, but the trend towards use of open equipment is clear.
On the network side, Facebook revealed it has significantly stepped up its use of Wedge and 6 Pack (its OCP chassis switch) from last year. The majority of new rack and data center sites being commissioned are using Wedge, which already provides connectivity for thousands of servers. These “light-house deployments are being watched very closely by data-center operators, as the trend towards open hardware with open source software continues to accelerate.
We can expect continued innovation that brings more diversity in compute and switching hardware and provides a choice of silicon. Strong open-source switch OS projects will foster additional innovation. The variety of bare-metal hardware and open-source software in the market continues to multiply and become more mature, providing alternate avenues for enterprises and CSPs to purchase and deploy data center infrastructure.