Fog-Dew Architecture for Better Consistency

3

July 2, 2016 by tushumane

Abstract:

Computing that is independent and collaborative with the cloud computing is referred as dew computing. The main principle that drives dew computing is the drawback of cloud computing, requirement of internet connection. Though dew computing will allow user to have functions and services offline, the duration for which it will be in inconsistent state with the cloud is important concern. The proposed technique in this paper introduces fog layer that will fill the time lag between dew and cloud services. An application is discussed later in the paper to  demonstrate the concept.

Keywords:

Dew Computing, Fog Computing, Cloud Computing, Consistency, Community Device, Application of Dew Computing

Introduction:

Cloud computing, one of the major evolutions in computer and networking field has proven to be most flexible and cost-effective solution to many problems [1]. With collaboration of virtualization it has opened doors to the world of computing where imaginations can come in to reality. Along with flexibility and cost effectiveness, some other benefits of cloud computing are control, disaster recovery , platform and device independency [2].  Among several limitations, high latency is big obstruction in use of cloud computing paradigm for time critical applications [3].  Most of the IoT (Internet of Things) applications are latency sensitive and require quick responses or services [4]. With this goal in mind Cisco introduced new computing paradigm, fog computing [5]. In Fog computing, services can be hosted at end devices such as set-top-boxes, access points or dedicated servers. This new distributed computing allows applications to run as close as possible to devices that generate huge amount of data or processes that produce data [6]. As Fog computing is implemented at the edge of the network, it provides low latency, location awareness, and improves quality-of-services (QoS) for streaming and real time applications. Example includes, healthcare, transportation, robotics, etc. Fog computing addresses the latency issue of cloud. But another major issue, requirement of internet connection to work was open. This lead to invention of another kind of computing architecture, cloud-dew architecture [7], where in minimal light weight local copy will be maintained on user’s device which will facilitate user to work even offline [8]. As internet connection is available, state of device will be synchronized with cloud and consistency is achieved [9].  A good example of this is Dropbox or google drive for desktop. Drive is always available, you can store, update or delete anytime you want, regardless of live internet connection. Still it is not full-fledged dew application, as it requires internet to keep track of versions. If version management function is provided along with other offline service, it would really be a complete dew computing application. But again this  leads to important issues like minimum acceptable time before dew application syncs with cloud to avoid further complex consistency issues. Complexity of consistency issues will grow exponentially with number of users in multiuser environment. As dew computing is in infant stage there may be many issues that need to be addressed, in this paper, attempt is made to resolve consistency issue to some extend with the help of introducing a fog layer in existing cloud-dew architecture. We have discussed two levels of architecture to support our ultimate objective: device level and application level. Device level focuses on the modifications to be done at device level to support dew computing while application level focuses on modification to be done at application level to achieve the same. These two approaches can be alternative to each other in some cases as well as they will complement each other in some other cases. It depends upon the context.

Device Level Architecture:

Figure 1: Overview of Architecture

Above figure shows three layers: user layer, community layer and cloud layer. User layer consists of user devices like personal computers, laptops or mobile devices. Laptop and mobile devices are connected to community server via wireless interfaces. Desktop may be connected to community server with serial interface or wireless interface. There is no need of active internet connection on user devices, and they are connected to community server on interface A with wi-fi or local area network connection. However, the other interface of community server, interface B, will be connected to internet. This will allow active synchronization with cloud servers.

Figure 2: Internals of End User Device and Interaction with Community Server

End user device will contain two isolated areas, data and control. These are jailed environments, and completely independent on private part, the operating system. Community server can access public part of device, it can push or pull data to/from data area, while control instructions (if any) will be communicated with the help of control area. Program or set of instructions will control behavior of data in data area. User can select what should be public. User can even put his own program in control area.

Device Level Application: Review System

Though the review system discussed below can be applicable to huge set of business areas, we will consider movie review system to demonstrate the concept.

Figure 3: Fog-Dew Based Movie Review System

As soon as individual enters in to movie theatre, he/she will get connected to open access community server.Community server will push an interactive review page in to public data area of user’s device. Notification instruction will be sent to control device which will be then shown to user like, “Welcome! Please feel free to rate movie and more “. Similar kind of notification will be sent to user in the interval and at the end of movie. Whenever user feels, during movie, during interval or at the end while going out of theatre, he/she can rate and review the movie, and response will be sent to community server. All responses will be collected and sent over the internet by community server. These reviews will be immediately available over internet, this has completely solved consistency issue in cloud-dew architecture. Again. instead pushing customized page, we can use Facebook or twitter to get live feeds but again, applications should be designed to support such architecture. We will discuss more about how applications might be developed  to support such architecture.

Application Level Architecture: 

We will discuss two scenarios to present above concept:

Dropbox: In case of active internet connection Dropbox will function normally, present case. As of now, when internet connection is not available, we can store, delete or update the files. That is final state of file is stored on cloud when internet connection becomes available and not the intermediate versions. So what if user seeks for intermediate state of file? To make it more dew supporting, dropbox may incorporate version management instructions which will work locally, on user’s device and not online. Once internet connection is on, current state and all the intermediate states, maintained by offline version management, will be saved on cloud.

Facebook Mobile Application: Let’s consider again our movie reviewing system. As soon as individual enters in movie theatre, community server will send the instruction to control area requesting interaction with Facebook. It will then push a copy of official facebook page of movie in to data area. Control area will glue that Facebook page pushed on data are to facebook app. User will not come to know about that the page is separate page, he/she will post a review on movie on that Facebook page, responses will be sent back to community server, which will extract the ratings/reviews from dummy page and will synchronize it with facebook server. Reviews/ratings will be available on internet immediately.

Conclusion:

With introduction of fog layer, inconsistency issues in cloud-dew architecture can be solved to some extent. Operating systems and applications may support dew computing by using device level and application level architectures respectively. Community server will definitely play crucial role in achieving offline computation. Architectures can be alternatives in some cases, while complimentary to each other in some other cases. Depending on requirements and context further modifications can be done. We had a bird’s-eye on architectures, further and deeper look may reveal some interesting facts and challenges.

References:

[1]  Cloud computing – concepts, architecture and challenges by Yashpalsinh Jadeja. Computing, Electronics and Electrical Technologies (ICCEET), 2012 International Conference on

[2]  Cloud computing: Need, enabling technology, architecture, advantages and challenges by Amit Gajbhiye. Confluence The Next Generation Information Technology Summit (Confluence), 2014 5th International Conference.

[3] Cloud Computing: A Survey on its limitations and Potential Solutions by Mohammad Manzurul Islam. IJCSI International Journal of Computer Science Issues, Vol. 10, Issue 4, No 2, July 2013.

[4] The Internet of Things – promise for the future? An introduction by Louis Coetzee. IST-Africa Conference Proceedings, 2011.

[5] Fog Computing and Its Role in the Internet of Things by Flavio Bonomi.http://www.ce.uniroma2.it/courses/sdcc1415/progetti

[6] The Fog Computing Paradigm: Scenarios and Security Issues by Ivan Stojmenovic. Proceedings of the 2014 Federated Conference on Computer Science and Information Systems pp. 1–8.

[7] Cloud-dew architecture by Yingwei Wang. Int. J. Cloud Computing, Vol. 4, No. 3, 2015.

[8] Definition and Categorization of Dew Computing by Yingwei Wang. Open Journal of Cloud Computing (OJCC), Volume 3, Issue 1, 2016.

[9] Cloud-dew architecture: realizing the potential of distributed database systems in unreliable networks by Yingwei Wang. Int’l Conf. Par. and Dist. Proc. Tech. and Appl., PDPTA’15.