Home > About Us > News Bulletins> Forrester Chief Analyst Danny Mu: Diversified Computing Leads the Intelligent Era

Forrester Chief Analyst Danny Mu: Diversified Computing Leads the Intelligent Era

2020-05-18

|The world is entering an intelligent age


Forrester recently conducted a study on diversified computing in which 338 executives from large and medium-sized companies around the world were interviewed. Respondents generally believe that the world is entering an intelligent era, which is reflected in three aspects:

 

Smart Applications Will Actively Enable Business Scenarios

 

Most respondents believe that intelligent data analysis will effectively improve customer experience and product design and R&D.

 

Intelligent Interconnection Will Positively Promote Industry Convergence

 

Respondents generally believe that the advent of the 5G era will greatly improve connectivity and lay a foundation for industry convergence.

 

Ubiquitous Intelligence Will Ultimately Help Enterprises Build Brand New Business Models

 

More than 90% of respondents want to use AI to create new digital products to expand into new business areas, and to innovate digital services through the use of a wide range of intelligent technologies.

 

|Conventional Computing Architectures Will Face Severe Challenges

 

At the advent of the intelligent era, we have witnessed explosive growth of intelligent applications. At the same time, Moore's Law becomes less effective. As a result, conventional computing architectures face  the following three challenges:

 

General-Purpose Computing processors Cannot be Used in Multiple Scenarios

 

For example,  with the use of processors in more intelligent scenarios, more powerful inference computing is needed. Another example is the Internet of Everything (IoE) in the intelligent era, it requires processors to cope with complex and extreme scenarios.

 

Difficult to Cope with Complex Environments with a Simple Computing Architecture

 

Nearly 80% of respondents believe that computing focuses on local data centers and public clouds, making it difficult to support explosive growth of intelligent scenarios. 75% of respondents think that the current computing architecture model is simple,  inflexible, and not adaptable.

 

Existing Computing Platforms Cannot Support Converged Innovation

 

A large number of respondents believe that deployment, O&M, and management of existing platforms are difficult. The development and testing process is too complex.

 

|Intelligent Era Requires Diversified Computing

 

To address these challenges, Forrester believes that diversified computing is required in the intelligent era. It can be applied to various application scenarios and provides ultimate computing power through heterogeneous processors, distributed architectures, and full-stack platforms.

 

Heterogeneous processors are Used to Provide Optimal Computing Power

 

 Since the data center is the distribution center of computing power,  its own intelligence and computing improvement are particularly important. Conventional data center computing mainly depends on CPUs, which cannot meet diversified service requirements and is not the most cost-effective solution. Therefore, we need to add intelligent features to reduce the CPU load.

 

 We can improve computing efficiency in data centers through AI. For example, intelligent acceleration engines and management engines. The intelligent acceleration engine improves server performance from the computing and storage network dimensions. It offloads some service applications to hardware, saving CPU resources, improving server performance, and transforming conventional data centers to intelligent data centers. Intelligent management software improves data center O&M efficiency.

 

These two intelligent engines break the CPU boundary, server boundary, and L1 and L2 boundary. Outside the data center, energy-efficient computing power is provided at the edge side, and customized computing power can be provided in many specific service scenarios.

 

Leverage Distributed Architecture to Fully Activate Business Scenarios

 

We not only need dedicated processors for diversified computing, but also integrated systems to fully release the computing power of specific processors.  Besides that, storage and networking capabilities also need to be matched. To meet the performance requirements of explosively growing intelligent applications, we must ensure that there are no shortcomings in the application process. Today, both storage and networking can be upgraded through computing in a software-defined manner. From the perspective of the breadth of intelligent applications, we  need a hyper-distributed architecture that links local data centers, public clouds, edge, and device. We also need a software-defined, flexible, adaptive, and reconfigurable architecture to support AI-enabled business scenarios.

 

Full-stack Platform Ensures Application Implementation

 

More and more complex systems require intelligent O&M and management due to diversified computing and hyper-distributed architecture. As the underlying layer uses diversified processors, we need one-stop development and testing capabilities when developing intelligent applications. The hyper-distributed architecture requires convenient application deployment capabilities. In addition, we need end-to-end security protection capabilities to maximize user privacy and enterprise information security.

 

We need to fully realize that intelligence is a complex process. Enterprises should not only actively embrace diversified computing, but also build strategies, cultures, and ecosystems in addition to the technology field to fully grab the opportunities and face the challenges brought by the intelligent era and better serve customers. 

636 Read
|
(0 comments)

Do you like this? Give your score

500/500

0 comments

Contact Us

Sales Hotline

+86 400-080-6888

Technical Support

+86 400-009-8999