Categories
Uncategorized

Edge computing: new computing model of all things Internet era

table of Contents

    EDITORIAL

    background

    Existing problems

    definition

    Advantage

  • 挑战

      Programmability

      Naming Rules

      Data Abstraction

      Service Management

      Data security and privacy protection

      Theoretical basis

      Business Model

  • Related research

    to sum up

    reference

EDITORIAL

This article is to read the paper “edge computing: all things Internet era of new computing model,” the notes.

background

At present, the data processing has been calculated from the cloud era centralized processing center (the 2005–2015 Nian centralized call processing big data era) is entered to all things Internet as the core edge computing era (called edge large data processing era). Large centralized data processing times, more of a centralized store and process large data, the way they have taken is to build a cloud computing center and cloud computing center super computing power to solve the centralized computing and storage problems. In contrast, at the edge of the era of big-data processing, network edge equipment generates vast amounts of data in real time; and, these edge devices will be deployed to support real-time data processing edge computing platform to provide users with a large number of services or functional interface, users can call these interfaces to obtain the desired edge computing services.

Existing problems

With the rapid development of the Internet of Things (IoT) and the spread of 4G/5G wireless networks, the Internet of Everything (IoE) has arrived, and the number of devices on the edge of the network has increased rapidly, bringing the data generated by such devices to the ZB level. In the era of centralized big data processing based on cloud computing models, the key technologies are no longer able to efficiently process data generated by edge devices. There are four main problems:

    The linear growth of centralized cloud computing power does not match the explosive growth of massive edge data;

    Mass from the transport network edge device to a cloud data center leading to a sharp increase in the loading of network bandwidth, network delays resulting in longer;

    Personal privacy data network edge, making privacy issues become particularly prominent;

    A data transmission network edge device power is limited to the center of the cloud of large power consumption.

Definitions

An edge calculating means calculating a new model calculation is performed at the network edge, the downstream data represents an edge calculation edge cloud services, Internet services all uplink data represents, and calculates an edge refers to an edge from a data source to the path of the center of the cloud any computing and network resources between.

1, the solid blue line represents the transmission data producer source data center to the cloud, solid red line represents the data consumer transmits a request to use the center of the cloud, the cloud red dotted line represents the results back to the central data consumer.

2 shows a calculation based on the edge Bidirectional flow model. Cloud computing center not only collect data from the database, but also to collect data from sensors and smart phones and other edge devices. These devices both data producers and consumers. Thus, the transmission request between the terminal apparatus and the center of the cloud are bidirectional. Only a request from a network edge device and the center of the cloud content services, but may also perform part of computing tasks, including data storage, processing, cache, device management, privacy protection.

Advantage

Migrating legacy edge computation model cloud computing center part or all of the tasks performed near the data source. 3V from the characteristics of large data, i.e. the amount of data (Volume), timeliness (Velocity), diversity (Variety), is calculated by comparing the cloud model representative of a large centralized data processing (FIG. 3) and the edge calculation large edge data processing model data representative of various characteristics of the times (FIG. 4) to illustrate the advantages of the edges of the calculation model.

challenge

7 key pressing issues that may be encountered in the study of edge computing, including: programmability, naming conventions, data abstraction, service management, data privacy and security, as well as the theoretical basis of the business model.

Programmability

One advantage of the cloud computing model application development infrastructure transparent to the user. User programs are usually written and compiled on the target platform, running on a cloud server. In edge computing model, some or all computing tasks need to migrate from the clouds to the edge node and edge nodes are mostly heterogeneous platforms, runtime environments on each node may vary, therefore, on the edge of the calculation model when deploying user applications, programmers will encounter greater difficulties. The existing traditional programming mode MapReduce, Spark, etc. are not suitable for an embodiment based on the study of new programming edge computing.

Naming Rules

In the edge calculation model, a great number of edge devices, and a computer system similar to the naming convention, the edge calculation naming a very important role in programming, addressing, identification and data communications, and more efficient No current naming rules. Edge computing naming rules not only meet the needs of communication between heterogeneous devices, mobile devices also need to meet, highly dynamic network topology, privacy and security requirements. The traditional naming mechanism such as DNS, URI meet most of the network structure, but not the flexibility to provide services to the dynamic edge of the network, because most of the edge device has a high degree of mobility and limited resources, and IP-based naming, because complexity and overhead too large to be applied to the edge calculation.

Data Abstraction

You will encounter three kinds of challenges edge computing data used in the model:

    Diversity of data format transmitted by different devices.

    Data abstraction degree of uncertainty. If the source data abstraction filtering more data, applications or services that will result in some programs due to inability to obtain sufficient information fails; on the contrary, if you keep a large number of data sources, data storage and management will be another challenge faced by system developers . Further, the data sent by the edge unreliability, how information from an unreliable source of useful information abstracted remains a technical challenge.

    Data abstraction applicability. Due to the heterogeneity of the edge device, and data representing the operation is also different, which will become the universal data abstraction barriers.

Service Management

Computing management service edge, any of a reliable system has four kinds of characteristics: the difference (differentiation), extensibility (Extensibility), isolation (Isolation) and reliability (reliability).

    Differences: edge of the network will deploy a variety of services, and these services need to have different priorities. Such as health-related services a higher priority than other entertainment services.

    Scalability: the need to design a flexible and scalable edge of the operating system for the management service layer.

    Isolation: Due to the presence of an edge model calculates a plurality of applications share the same data, when an application crashes not affect the other needs of the application accessing the data. In addition third-party programs and how to isolate the user’s private data is also a challenge.

    Reliability: From a service point of view, when the service to be able to quickly diagnose the failure or recovery services; from a system point of view, can be deployed at the system level fault detection, equipment replacement, and quality inspection data services; data from point of view, to be capable of data communication unreliable, reliable service.

Data security and privacy protection

Data privacy protection and security is an important service provided by the edge of computing. Compared to all the data processing in the cloud computing center, in the vicinity of the source computing data privacy and data security is the protection of an effective method.

Theoretical basis

Edge computing the theoretical basis of the current is not mature, integrated computing needs, there are relatively sound theoretical basis for data communication, storage and optimization of energy consumption and other disciplines, presented a comprehensive multi-dimensional edge computing theory, which is currently carried out at the edge key primary problem of computing research. Reasonable calculation of the theoretical basis of the edge has a very important significance for the future of academia and industry to better carry out applied research and development services computational model based on the edge.

Business Model

Computing across the edge information technology (the IT), communication technologies (CT) and other fields, to a plurality of industrial hardware and software platform, network connection, data aggregation, chips, sensors, and other industrial applications role chain. Edge computing business model will not only are more service-driven and user requests the corresponding service, and more will be data-driven.

Related research

With the development of big data era, in order to solve the problem of cloud computing and data center computing load transmission bandwidth, researchers have proposed a variety of techniques for computing tasks migrate from a cloud computing center to the edge of the network, which mainly typical model includes: distribution database model, P2P model, models the CDN, the edge of the mobile computing model, fog and Haiyun calculation model calculation.

to sum up

Edge computing model is not either-or relationship with the cloud computing model, but complementary relationship, the edge of big-data processing era is the era of mutually binding edge computing model and the cloud computing model, organic combination of the two will be all things internet era of information processing to provide a more perfect hardware and software support platform.

reference

Shi Wei Song, Sun Hui, Cao Jie, and so on. Edge computing: all things Internet era of new computing model [J]. Computer Research and Development, 2017,54 (5): 904-924.

Leave a Reply