OpenStack Day Taiwan 2018 - a major annual event for the Taiwan open source community - took on a whole new look with OpenStack renamed OpenInfra. Carrying on with the same OpenStack spirit and architecture, OpenInfra will engage more partners specializing in open source technologies and continue to flourish in the world of cloud computing.
Themed on the ingenuity of open source platforms and innovations combining different technologies, OpenInfra Day Taiwan 2018 highlighted topics including multicloud architecture and its deployment, container development, distributed data storage and software-defined everything. The event also presented use case examples in applications such as artificial intelligence (AI), edge computing and 5G IoT.
Delivering an opening speech at the event, OpenInfra Day Taiwan 2018 organizing committee chairman Brian Chen remarked that this year's event gathered influential leaders to share their views on the latest OpenInfra trends. The change in the organization's name allowed it to expand the discussion topics and engage more industry ecosystem partners. Providers of private, public and hybrid cloud IT infrastructure and services were invited for discussions on multicloud architecture and its deployment. For Infra, 5G and IoT, the participants looked into use case examples of edge computing, combination of 5G and IoT, as well as developments of IoT devices and application services. The discussions related to software-defined everything focused on the trends of software-defined storage, network and infrastructure. Popular AI topics including deep learning and machine learning were also highlighted at the event this year. In closing, Chen hoped the event would load participants full of inspirations and help advance Taiwan's IT and open source developments.
A close look at current developments of OpenStack with a market value of US$6.1 billion
Jonathan Bryce, OpenStack Foundation executive director, gave an account of OpenStack's development history and the reason for the name change to OpenInfra. With user groups consolidating and open source software growingly used in vertical applications, open source infrastructure actually encompasses a wider scope than OpenStack, so the name change to OpenInfra represents open source code, open source design, open source development and open source community, making open source software infrastructure available to more businesses.
Bryce continued to present some important data. According to the results of the foundation's user survey, OpenStack is supporting infrastructure running across more than 10 million CPU cores around the world. Among OpenStack users worldwide, 71% are corporate users deploying OpenStack in production environments or are preparing to do so within the next 12 months. Lastly but most importantly, the OpenStack market size is estimated to top US$6.1 billion in 2018.
Applications today have different requirements for OpenStack, cloud and infrastructure than before. They need AI machine learning, container, ML and serverless technologies more than ever before, which drives rising demand for open source software for use outside of datacenters by edge computing, added Bryce.
According to Bryce, OpenStack Foundation has realized open source is not just transitioning from open source code to open source infrastructure but more importantly, it should build an open infrastructure community, where humans are the most critical element. As such, the migration from OpenStack to OpenInfra represents a change from the perspective of infrastructure to the perspective of community.
The foundation is transcending the open source spirit to a higher level with the transition from OpenStack to OpenInfra. Open source software will become not only omnipresent but also part of people's daily life in the future, summarized Bryce.
Continuing cloud native computing development focuses on openness, ecosystem collaborations and innovations
Cloud Native Computing Foundation (CNCF) was established for the purpose of promoting fast-growing open source projects on Github. CNCF China director Keith Chan shed light on the current development of cloud native computing and CNCF's achievements with respect to openness, ecosystem collaborations and innovations.
Chan pointed out the open source model has played an important role in global trade and business models. The next step to promoting the open source model is for businesses to form an ecosystem.
In regard to innovations, Chan introduced CNCF's organizational structure and functions: Technical Oversight Committee mainly approves new projects; End User Community serves to help CNCF gain a closer understanding of user needs and make open source development more in line with market applications; Marketing Committee aims to match promotion efforts, for example, for Kubernetes, closely with market dynamics. CNCF is also offering conformance certification for software based on Kubernetes, with which software vendors can display or make flexible use of the Certified Kubernetes logo. CNCF will continue to engage contributions from cloud native computing experts worldwide.
Current development of Kata container and comparison to gVisor
Xu Wang, CTO and co-founder of HyperHQ, introduced the company's open source Hyper, which can run Docker Image directly on Hypervisor, as well as Kata container, an open source container infrastructure project managed by the OpenStack Foundation.
According to Wang, Kata container was developed with the goal to make it independent of hardware architectures, so it not only is compatible with the OCI runtime specification but also supports the Kubernetes Container Runtime Interface (CRI).
Wang also shared his observation on a special development – information security concern about Kata container becoming the focus of attention. Then, he gave a comparison between Kata container and gVisor. gVisor is a new sandboxed container runtime authored by Google, which provides secure isolation for containers while being less resource intensive than a full virtual machine. In comparison to Kata container, gVisor supports most Linux system calls and is designed with different trade-offs than existing container technology. Wang wrapped up his talk with a few key characteristics about Kata container. Like runC containers, Kata containers deliver the security of virtual machines while supporting other VM accelerating technologies. Despite using virtualization, Kata containers are light, fast and easy to integrate into many different application workflows. It supports multiple architectures including x86_64, arm64 and ppc64le.
Large-scale AI cloud platform architecture and service, and NCHC's five major strategies for AI development
Dr. Feng-Yi Chao of National Center for High-Performance Computing (NCHC) gave a talk, themed on large-scale AI cloud platform architecture and service, about how NCHC helps Taiwan establish five major strategies for national level AI development.
NCHC's Taiwan GPU Cloud currently serves 66 projects undertaken by five major AI research centers and 13 subprojects under the Semiconductor Moonshot Project by providing users a single container platform with single authorization. For the long-term objective, Chao said NCHC aims to offer enterprises SLA-based AI-HPC services, build a demo project with NVIDIA GPU Cloud (NGC) in Taiwan and provide Data Mart services, social media hosting services as well as Learn to Learn AI services.
In summary, Chao shed light on NCHC's role at the international level. Focusing on AI-HPC cloud services, the goal is to provide a selection of flexible service models where CPU, GPU and other public-cloud-based computing services are consolidated through cloud resource management software. It not only effectively increases the usage rate of domestic AI computing platforms but also combines public and private clouds to carry out computation for researchers on cloud platforms worldwide through a single management interface with uniform features in one implementation.
As the Taiwan government is devoting tremendous resources to fostering AI developments, spurring burgeoning innovations, now is a good time for AI undertakings. Keeping a clear vision of the goal and the strategy to find partners, build teams and make alliances, NCHC endeavors to build hardware and software platforms for use by the industry, government, research institutions and academia.
Innovative AIoT applications are the popular trend, spurring edge computing developments and opportunities
AIoT innovations are grabbing market attention in recent years. During early development in 2013, IoT applications mostly targeted smart grid and home surveillance with low-volume data collection. Today's IoT applications process large amounts of data. Their software and hardware are growingly incorporating AI capabilities, giving rise to widespread implementations of edge computing. According to Norr Wei, industry analyst, Industry, Science and Technology International Strategy Center (ISTI) at Industrial Technology Research Institute (ITRI), the rise of edge computing has much to do with AI + IoT applications. In some AI use scenarios, cloud computing architectures are beginning to encounter challenges – limited bandwidth, communication latency, data security and confidentiality concerns, and inadequate network coverage among others.
For its low data latency and high transmission frequency, edge computing will enjoy rapidly rising demand particularly from IoV, AR and VR applications. It is estimated 75% of data will be processed and even stored at the edge by 2022.
According to ISTI, solutions targeting vertical applications will present better opportunities for vendors to penetrate into the edge computing market. As to the business models, B2B IoT applications are generating rapidly growing values while the B2C segment will focus on smart home applications with service providers like Apple, Amazon and Xiaomi dominating the market. The competitiveness of edge computing will hinge on whether it can achieve economies of scale. Cloud service providers are betting on increasing their volume of connected platforms. IT hardware suppliers are looking to expand the deployment of their products through vertical applications and cross-sector collaborations.
To Taiwan-based vendors expanding into edge computing, Wei suggested two development strategies. First, they should form strategic alliances with international platform operators to jointly develop solutions integrating software and hardware or make their way into the supply chain. Second, they can focus efforts on applications with tremendous potential.
Four separate sessions focusing respectively on multicloud, IoT, software and AI
The afternoon sessions of OpenStack Day Taiwan, respectively themed on diverse topics including multicloud architecture and its deployment; Infra, 5G & IoT; software-defined everything; AI, presented real use cases and advanced technologies, fully demonstrating the OpenInfra spirit.
OpenInfra Day Taiwan 2018
DIGITIMES' editorial team was not involved in the creation or production of this content. Companies looking to contribute commercial news or press releases are welcome to contact us.