Encyclopedia of Wireless Networks

Living Edition
| Editors: Xuemin (Sherman) Shen, Xiaodong Lin, Kuan Zhang

Data-Driven Service Provisioning

  • Huawei Huang
  • Song Guo
Living reference work entry
DOI: https://doi.org/10.1007/978-3-319-32903-1_93-1
  • 40 Downloads

Synonyms

Definitions

Service provisioning is the process of configuring a service for customers. It usually associates with telecom industry and cloud infrastructure. The examples of service provisioning are shown as follows.
  • When a customer submits a request to use a mobile network, system initializes and manages the customer’s services, such as establishing control channels (Huang et al., 2016) and making update scheme for service provisioning due to user mobility (Huang and Guo, 2017).

  • To activate a low-earth-orbit satellite-based service for secure big data storage, the telecom company has to design and construct the infrastructure for customers in advance (Huang et al., 2018).

  • Data centers or edge nodes launch a Service Function Chaining enabled machine to meet the customer’s subscription, including the desired software and hardware resources, and the access authority in the agreed period (Huang et al., 2017b).

  • Cloud infrastructure provider maintains a pool of instances. Customers are allowed to request these instances through application programming interfaces (API) (Xu et al., 2015; Huang et al., 2017a).

Background

Big data promises to reshape the knowledge, society, and economy. The technologies that are used to generate, collect, process, compute, and communicate big data have made significant progresses in the past few years. As an overwhelming volume of data is generated with ever-growing speed every day from various sources and applications, such as cloud services, Internet of Things (IoT), social network services, and other smart terminals, it becomes significant to design, deploy, and provision services that are more wise to enable the effective acquisition, storage, transformation, management, and utilization of such huge amount of data.

To alleviate the violations of service-level objectives caused by provisioning delays, inefficient reconfigurations, and inaccurate demand predictions, over-provisioning apparently increases operational costs and resource-utilization inefficiencies, particularly in large-scale systems. Via a data-driven service provisioning framework named AidOps, Lugones et al. (2017) argue that the orchestration system enables a higher service availability and lower over-provisioning costs.

Foundations

As mentioned by Chesbrough and Spohrer (2006), the growing knowledge of information and communication technology (ICT) has created opportunities to inspire the evolution of internet on the service relationships that can yield new value. Specifically, ICT has the following characteristics that can improve effectiveness and efficiency and enable innovativeness in the manners of (1) commoditizing noncore competency such as outsourcing, (2) improving collaboration in business processes, (3) controlling the risks of information security violation, (4) coordinating service systems such as open innovation platforms, (5) improving customer-oriented service quality, etc. Especially, in the big data era, enormous value can be extracted from the service-oriented big data.

To fully explore the value of service-generated big data, developing efficient technologies that can fast process large amount of data is an open issue. On the other hand, easy access of big data and data analysis results are critical to both service providers and service users (Zheng et al., 2013).

Key Applications

In the following, some typical data-driven platforms/architectures for service provisioning are reviewed.

To provide the functionality of managing and analyzing service-generated big data, Zheng et al. (2013) proposed an infrastructure of data-driven service model. As shown in Fig. 1, a big data-as-a-service framework has been also employed to provide big data services and data analytics results to users. Under this framework, several concrete applications that exploit three types of service-generated big data are illustrating for performance enhancement. First, the service logs generated by service requests can be used to yield visualized log and conduct performance diagnosis. Second, based on the QoS information of each service, fault tolerance and QoS prediction can be performed by service providers. Finally, taking the service relationship into account, significant service identification and service migration are able to realize.
Fig. 1

Service-generated big data and big data-as-a-service as presented by Zheng et al. (2013)

A framework for service-oriented decision support systems (DSS) has been investigated by Demirkan and Delen (2013). By focusing on the product-oriented environment and exploring engineering-related issues, a conceptual architecture of service-oriented DSS is shown in Fig. 2, which includes three service models, i.e., data-as-a-service, information-as-a-service, and analytics-as-a-service. In such a service-oriented DSS solution, the operational systems, data warehouses, online analytics processing, and end-user components can be provided to users as services either separately or in the bundled manner.
Fig. 2

A conceptual architecture of service-oriented decision support systems presented by Demirkan and Delen (2013)

As mentioned previously, a systematic approach named AidOps for capacity provisioning of services in cloud is proposed by Lugones et al. (2017). Instead of on-demand elasticity, AidOps enables service providers to specify a desirable number of reconfigurations (i.e., scaling in/out) as an input, and it then calculates the capacity configurations required to serve the user demand optimally. Figure 3 illustrates the architecture design of AidOps, including system components and interfaces to the cloud and service layers. This system is consisted of a backend and frontend. The former supports data analytics and service provisioning algorithms, while the latter offers configuration functionality to operators. Specially, in the backend, the historical traffic data from operator database is preprocessed to the proper time series format and then fed to the clustering and classification modules for extracting workload patterns. With the pattern database, AidOps predicts daily patterns using a pattern selection module. Finally the predicted time-varying patterns are used to optimize the configuration solution that is demanded by service orchestrator.
Fig. 3

AidOps architecture proposed by Lugones et al. (2017)

Facing the convergence of IoT and cloud, it is significant to facilitate the service provisioning and management in large-scale applications, particularly under the scenario of smart cities. To model and expose cloud-based IoT services in an efficient manner, Taherkordi et al. (2018) proposed CARIoT as shown in Fig. 4, which is a software framework for real-time provisioning of cloud-based IoT services. The main idea behind CARIoT is to structure the description of data-driven IoT services and the service-generated data, both real-time and historical, in a hierarchical perspective corresponding to the context model of end-user application. With CARIoT, the end-user applications thus can access IoT services and their data in an efficient way.
Fig. 4

An overview of the IoT service model in cloud platforms, as presented by Taherkordi et al. (2018)

Important Issues

The important issues under the data-driven service provisioning can be summarized as but not limited to the following topics.
  • Efficient models and solutions for querying and processing big data.

  • Big data-based service modeling, delivery, deployment, and evolution.

  • Analytic services for big data.

  • Knowledge discovery over massive datasets.

  • Social sensing and social network services.

  • Reasoning over uncertain data and unreliable services.

  • Big data based service ranking and recommendation.

Cross-References

References

  1. Chesbrough H, Spohrer J (2006) A research manifesto for services science. Commun ACM 49(7):35–40CrossRefGoogle Scholar
  2. Demirkan H, Delen D (2013) Leveraging the capabilities of service-oriented decision support systems: putting analytics and big data in cloud. Decis Support Syst 55(1):412–421CrossRefGoogle Scholar
  3. Huang H, Guo S (2017) Service provisioning update scheme for mobile application users in a cloudlet network. In: IEEE international conference on communications (ICC). IEEE, pp 1–6Google Scholar
  4. Huang H, Guo S, Liang W, Li K, Ye B, Zhuang W (2016) Near-optimal routing protection for in-band software-defined heterogeneous networks. IEEE J Sel Areas Commun 34(11):2918–2934CrossRefGoogle Scholar
  5. Huang H, Guo S, Wu J, Li J (2017a) Service chaining for hybrid network function. IEEE Trans Cloud Comput,  https://doi.org/10.1109/TCC.2017.2721401 CrossRefGoogle Scholar
  6. Huang H, Li P, Guo S, Liang W, Wang K (2017b) Near-optimal deployment of service chains by exploiting correlations between network functions. IEEE Trans Cloud Comput,  https://doi.org/10.1109/TCC.2017.2780165 CrossRefGoogle Scholar
  7. Huang H, Guo S, Wang K (2018) Envisioned wireless big data storage for low-earth-orbit satellite-based cloud. IEEE Wirel Commun 25(1):26–31CrossRefGoogle Scholar
  8. Lugones D, Aroca JA, Jin Y, Sala A, Hilt V (2017) Aidops: a data-driven provisioning of high-availability services in cloud. In: Proceedings of the 2017 symposium on cloud computing. ACM, pp 466–478Google Scholar
  9. Taherkordi A, Eliassen F, Mcdonald M, Horn G (2018) Context-driven and real-time provisioning of data-centric IoT services in the cloud. ACM Trans Internet Technol (TOIT) 1(1):1–20CrossRefGoogle Scholar
  10. Xu Z, Liang W, Xu W, Jia M, Guo S (2015) Capacitated cloudlet placements in wireless metropolitan area networks. In: IEEE 40th conference on local computer networks (LCN). IEEE, pp 570–578Google Scholar
  11. Zheng Z, Zhu J, Lyu MR (2013) Service-generated big data and big data-as-a-service: an overview. In: 2013 IEEE international congress on big data (BigData Congress). IEEE, pp 403–410Google Scholar

Copyright information

© Springer Nature Switzerland AG 2019

Authors and Affiliations

  • Huawei Huang
    • 1
  • Song Guo
    • 2
  1. 1.School of Data and Computer ScienceSun Yat-sen UniversityGuang zhouChina
  2. 2.Department of ComputingThe Hong Kong Polytechnic UniversityKowloonHong Kong

Section editors and affiliations

  • Song Guo
    • 1
  1. 1.Hong Kong Polytechnic UniversityKowloonHong Kong