Fog Computing-I

hello so we'll be continuing our discussion on cloud computing so what we have seen that in case of cloud computing what we are trying to do we are trying to offload our computing and computing processes and data to the cloud right so that it is maintained by a third party and the on the other end the customer or the consumer or the user more concentrate on the business processes process so that is the basic objective of or model of the things right there are there are a lot of technical technicalities at the backend but nevertheless we are offloading the thing so what for that what we need a very strong backbone or a strong backbone Network which should be always up and and able to transfer data on a large volume as we see as as as along with the development and being most of the things are digitally enabled we are what we are getting a huge volume of data in other sense a huge volume of data maybe need to be transmitted from this customer end or consumer end to this cloud service provider being executed the results in some cases are transmitted back or transmitted in other places right the major issue is this huge volume or transfer of data and what we have what we see in recent development with number of activities specially Internet of Things coming up and more only and huge variety of sensors in place so we have lot of multimedia data which need to be transmitted and that is one part of the story that we require a huge backbone and type of things as far as the cloud is concerned what we consider it's a it's a huge computing power much higher than what we what the devices can do and it is some sort of infinite computing power is there on the other hand what we see that a huge volume of data are being generated and being transmitted over the network in what we see that the devices starting from as we discussed about mobile devices smart mobile devices or other type of diseases or even intermediate network devices what they are becoming is more more powerful in terms of computation and more resourceful things or in other sense we are not all the times exploiting the resources available in the things SiC say consider a particular sense sensor node or a local sync node of a sensor which are collecting the data and transmitting to the in the upward path maybe to the cloud so this this could have been done some processing at time like I can say that suppose in this particular room or a particular lab I have say 10 temperature sensors right so what what what what is my basically business model this temperature should be varying between say 18 to 22 degree centigrade that is the that is that operating range of this temperature now what we are doing this all these T naught sensors are sending the data to the up in the server maybe in a cloud that is calculating whether the within the limit and type of things and I can have say 10 such labs so there are 100 such data are going on and if the temperature is varying somewhere it is sending error now if you consider this particular a a particular a enclosed lab or single things otherwise I have what I have could take a night in local DS and whether my temperature is okay in the lab by the by the sink node of the sensors which are collecting this data of this particular room and it takes a call that whether higher or upper and then sense that say it's some statistical data or what we say some aggregated data to the sensor it may be the average or it may be average with other standard deviation etc to the other things in other sense it is not enough in other than sensing this transmitting the stain since as data I am sending one average data or and which has my purpose even you can say that if the if my sink node is intelligent enough it can take a call that whether the temperature is within this operating range yes or outside thing some zero one or yes/no type of things and then transmit this in other sense this is taking some part of computing ethics so with the intermediate devices becoming more intelligent whether there is a possibility of pushing the computing from logically centralized cloud to somewhere more down the line right towards the edge of the things right that's exactly what we are trying to discuss today is what we say this sort of computing is fog or from cloud to fog right so cloud is the whole thing and for computing so as we see the challenges or the the data what the cloud computing today's is doing the processing of huge data in the data centers data centers may be privately hosted or publicly available by paying rain that is it can be a public cloud or a private cloud all necessary information has to be uploaded or transmitted to the cloud for processing and extracting knowledge of it right so the whole data as we are discussing need to be transmitted to the cloud now also we have seen the typical characteristics of cloud for WA for which we we are the today's world is inclined towards is that dynamic scalability I can scale up or scale down based on my need so another is that no infrastructure management or practically minimal infrastructure management at the user in so if I load everything that computing it set down the cloud so I require very less infrastructure management at my user in and secondly and finally what we have a metered service right pay-as-you-go model so these three things that dynamic scalability minimal management or all my infrastructure management pushing it to the cloud and metered service pay-as-you-go model these are primary features of the cloud which makes is popular there are several other things which are which are there but never delays these are the three things which are the driving force so whatever we do we do not want to lose out of the things if we compromise on those type of features then the very motivation to going towards cloud may be challenged now there are issues with cloud only computing so what we say that only the cloud is computing register sitting duck maybe the issues especially in today's applications which variety of sensors variety of real-time operations and lot of redundant data right there are lot of data which are redundant like if I am sending temperature things it may not mean may be meaningless to sense all the sensors data which are which are more or less same information unless there is a different in some sensor data I may not want to send that it all are reporting between around 20 degree centigrade it does not require a cloud to take a call it could have been done as a much lower level so that or in other sense I have a huge amount of digital data to be transmitted so communication takes place takes a long time due to human smart for interaction and type of things if the state still data centers are centralized right data centers and in some Woodson so all the data from different region can cause congestion in the core right so being transmitted things especially in case of exigencies where a lot of volumes of data suddenly pushed into the thing right in case of say some disaster or some huge amount of in flux due to some event this is a lot of volume of data suddenly in flux there is a huge volume of data to be transmitted and there can be congestion and such a task requires very low response time to prevent further kasev Center so if I have this sort of things which has a some sort of accident some accident prevention mechanisms can into should be activated so where we require a very low response time so immediately need to be act acted so waiting for that cloud to take a call revert back and all those things may take lot of time so that is another problem so so the emergence of a concept called for computing so on the cloud we are – we are talking about fog that is little bit bringing Tao down to the ground or in other sense we are pushing this computing thing from the from the centralized data center or the cloud data centers to this edges right or intermediate or the edges of the network AJ age of the network so for computing also known as fogging and edge computing though some people have little other views of that edge computing but nevertheless it is a fogging or H computing it is a model in which data process applications are concentrated in devices at the network age rather than existing almost entirely on the cloud so now not only the cloud at the centralized things the data application and processes are distributed in the H right which which some effect of some way of distributing this whole processing whether things what it helps us it helps us in reducing the data load in the communication I can have a local decision and which is not needed for the global type of things they say smart traffic light management system the traffic light management system in Kolkata is nothing to do with the traffic light management system in Delhi apparently right for day-to-day traffic management right so I could have done it locally or even I can say that a region of a particular city may may have only aggregated data which need to be transmitted at the higher level for traffic management right so that basic intermediate management could be done locally so those things could be done in a concept of what we say fogging or for computing the term for computing was first introduced by Cisco as a new model to ease wireless data transfer to distributed devices in the Internet of Things Network paradigm so as IOT is becoming omnipresent or IOT is becoming a everywhere is there Internet of Things so it's huge volume of data devices which mass computing capability or resources much higher resources can do a bit of a job which could have been solved at a at a lower level so Cisco's vision if we look at that for computing is to enable application on billions of connected devices to run directly on the network edge since Cisco is primarily a network driven organization so it has a huge number of devices across the world and those devices are somewhat managed etcetera managed by a some sort of a Magina T is therefore because upon the one make and there are resourceful devices which could have done some sort of computing things and I can even run applications on the devices and doing so and so forth right so user can develop manage around software application of Cisco framework of network devices including HUD harden routers switches etc cisco brings say open source Linux and network operating system together in a single network devices so it helped to do not only computing but the first if you want to do computing you need to give some sort of a platform to run the applications for the computing things right so those things are they are in the devices and this this this is possible because of there are resources available at different layer of the network towards the edge so if we look at a view so this cloud ur at the top it is still there and it should be there there are intermediate devices which we are now helping only so far was only transmitting the data now can they do some sort of a computing what we say for computing and there are end user devices which are spread over different locations starting form say smart vehicles or which can communicate devices servers smart cameras and anything which can do write any any any any device which can capture detailed data compute and transmit right so bringing intelligence down from cloud closer to the end-user or the edge of the network that is one of the thing cellular base station network routers Wi-Fi gateways will be capable of running these applications right so there are because whenever I communication we have cellular networks Wi-Fi router into place and if those are having surplus resources and they are able to do that so that I mean my application can run say I want to run a application for monitoring the environment of different labs starting from temperature to humidity may be some sort of a what sort of air pollution or air content etcetera so this sort of things can be done in devices like sensors are able to perform basic data processing right so the sensors can basic data processing processing close to the devices lowers the response time enabling real-time applications right so whenever we process close to the devices so the response time reduces that is obvious and I can do lot of real-time processing of the things right so I can do a real-time processing of a of a say of any applications like I do a application based on that what we say dynamic signaling mechanism of a traffic light based on the traffic on the road so the cameras which are on the road capturing that how many what is the traffic flow based on that I the traffic nicing may change if that is the that is the need of this traffic management so that is local right local to a particular portion local to a region local to a city right so that definition of locality may vary from application to application but what we require that your devices like the traffic light device etcetera should be able to run this application which can take a call right so those are things nevertheless this is this is about the form so if we look at for computing enable some transactions and resources at the edge of the cloud rather than establishing channels for the cloud storage and you utilization so rather than just transmitting it do some sort of a transaction processing or application running on the things for computing reserves reduces the need of bandwidth by not sending every bit of information to the cloud channels over the cloud channel instead aggregating at a certain axis point so it agree gates and send the aggregate data this kind of distributed strategy may help in lowering cost and improve efficiency so this sort of it is a distributed strategy and this type of distributed phenomena may help us in lowering the overall cost not only in terms of monetary it the cost of transmission in terms of time etcetera and I can we can do efficiency later I can I can run several applications which can be real-time and type of things so this motivation is obvious already whatever we have discussed the motivation the for computing a paradigm that extends cloud and its services to the edge of the network for provides data compute storage application services to the end-user if you see it says some sort of a small form of a instance of the cloud for that local type of things right so it's it's doing some sort of a computing or giving some sort of a cloud service at that time at at that portion of that region descent and we and there is another side of the things because we have several series of developments one is the smart grid other is the smart traffic lighting in cities specially cities connected vehicles are strong regular networks which is coming up and also the software-defined network so these are the different aspects which are itself is a topic to work at but the smart grid smart traffic lighting smut vehicles software-defined network and so and so forth they are becoming pretty popular and in turn they generate huge volume of data right everyone is generating huge volume of data which are being transmitted at the higher up in the layer for doing that so all this different aspects has motivated or what a what it has pushed the push the processing towards doing it at the edges or intermediate layer rather than causing everything to the cloud so this is this is what we look at the form so this is the same thing what we discussed so we have one in this cloud so which has a data center with huge capability massive parallel data processing big demand big data mining machine learning algorithms etc so which is they are and should be their intermediate layer which is more near to this age or the devices so they are can act as a fork so they are they can be there can be fox sites with real-time data processing data caching computation of offloading and those type of things so these are not so powerful at that but as such they are intermediate devices we are which are used for transmitting data so that these are this can be used at the at the end or the at the front end or the age or the last mind what we say what we have the sensors we are connecting different type of data perform data pre-processing and compression mobile device serve as a human computer interfaces like this these are the different type of things which are transmitting out here and in turn transmitting to the things so these some sort of communication yes if we see that both way arrow can be taken a call at this end itself right without transmitting the whole data at the things it may be some sort of aggregated reporting and type of things or aggregating the data and taking putting it to the cloud for running some intelligent algorithm and machine learning based algorithm and type of things so we have more interactive and more responsive and to more computing power and more storage and at the other end so instead of just putting a channel to transmit everything to the cloud and compute and come back we are doing some intermittent the provisioning of intermediate processing for to serve the application based so this is this is definitely a major motivation and what we try to look at that has as we have seen that the typical properties of cloud that having here infinite scalability theoretical your quote unquote infinite scalability or of loading or having infrastructure no need of maintaining infrastructure at the client date or meter services those need to be need to be preserved or respected right in case of a form and those are definitely are still there as what we are discussed so there are for computing there are several enablers as those are true for our cloud computing also one is that virtualization so virtual machines can be used as the edge devices right so there are there can be virtual machines containers or containers services reduces the overhead of resource management by using lightweight virtualization or what we say container based application or services one of the popular container is docker container right so it the idea is it docks into that particular things and run on the thing so you do not have to that dependencies is carries along with the thing right so it's a again a separate topic if possible we will discuss sometime but that is this docking or container services are becoming very popular so that is another enabling technology out here such piece oriented architecture as we have discussed which is a enabling technology for cloud also is here also that SOA is a style of software design where services are provided to the other components by application component component through a communication protocol another thing so you have a service-oriented architecture which three major component of service provider service comma consumer and service registry so so that heterogeneous loosely coupled parties can talk to each other right so in so a architecture is one of the driving enabling technology and also what we are looking seeing at is the software-defined network right Sdn so Sdn is an approach of using open protocols like for example open flow to apply globally aware software control at the edges of the network to access the to excrete cheese routers that are typically would use closed and proprietary from one so this is this is another technology which which which is becoming pretty popular or already popular in software-defined network and which which is a which is enabling technology for our four four walls so with this several enabling technology form is becoming a reality and being deployed and used in several cases so in looking at so we should not see that fog as a replacement of cloud it is not a replacement of count not neither a competitor in that sense right it is basically offloading some of these workload from the cloud to this edge devices because the resources are available because there are applications which are real-time and needs more more quick responses and overall process may be cost-effective and efficient so for edge devices are there to help cloud data centers to better response time for real-time applications right handshaking amount fog and cloud is needed appropriate handshaking or synchronization between this fog and cloud is very much needed broadly benefits of computing can be that low latency and location awareness so it is aware that which location is operating widespread geographical distribution especially with the sensors etc so it has I think mobility there is another important thing like nowadays devices are we have lot of mobile devices right so the distance from the cloud or the intermediate devices which which say in device passing through the intermediate devices will change based on the mobility of the things now this requires a resynchronization reestablishment of the path had it been in locally somewhere it may is the computing and response time so Lowell attention location-aware net y-space this mobility very large number of nodes with as we are discussing with sensors and things predominant role of wireless access right huge volume of LS accesses strong presence of streaming and real-time application so these days we are having a huge streaming and real-time applications and which requires quick response time huge volume of data and type of things need to be processed quickly and may not require all data to be transmitted right so this huge volume of data can be locally processed and aggregated data can be transmitted so that the overall response time improves zero considerable way strong presence of streaming and return and heterogeneity different sort of devices different type of the mix and inner nd heterogeneous so I can have it some sort of a form sort of intermediate framework which basically talked to some devices which may be different from other devices like I have a group of sensors I have a their sink node which talks to the sensor also do this aggregation which may be different and another other set of sensors which has a different sink note but nevertheless when they do this aggregated data that is in a more standardized format so I can handle heterogeneous devices so advantage is already we have already discussed so can be distinguished from cloud by proximity to the end-user that is one of the advantage or over this free service cloud dense geographical distribution and its support for mobility right so we can have instead of scintillate I can have a lot of distribution it provides low latency low awareness and improves quality of service and real-time applications right so there is a there is a chance of better performing the things rather we try to look at it is not isolated form but for plot cloud o as a whole can give a better service to these consumers right in terms of cost in terms of scalability in terms of your efficiency and type of thing specially applications where we have high quality of services and real-time services streaming videos and type of things there are of course some issues related to security so one is that as devices are dispersed right so maintainability of the security protocols at different for devices is a serious challenge right so it is at different location now had it been cloud you have a provider at a particular centralized things you can put lot of security mechanism in the place but if you once you distribute over the form then you have to maintain so many things on the on different age devices so it is not only the data processing etcetera so there can be security issues so and many in the middle attack type of things can happen so as devices are dis parts differ as as this computing data are being there in the in different age devices so there is a issue of things of men in the middle letter can be there there are issues of privacy issues as same that it is it is being processed at different ages and whether the data leakage is there then whether you know about the things like if I consider smart grid or connected vehicles so if you do the intermediate port is processing whether you are basically tracking the vehicle or looking at the processing of the consumption of individual house or whom and type of utilization those can be there like in case of a smart grid smart meter installed at the consumer home a smart meter and smart appliance as an IP address a Malaysia has users can either tamper with its own smart meter report false reading or spoof IP addresses and so and so forth so whatever it comes with typical network security related issues may also come may also be problem out here so may be a challenge so there are definitely security issues there are security issues in the cloud but this extend that to much more things as you have different devices are activated so what we see today that this form is just not extension of the cloud it's a it's a necessity based on the different application huge volume of data and the devices intermediate devices becoming more resourceful right and they are able to capable to do this type of calculations computation and secondly in order to in doing so I may not be doing all these high profile computation but I can we can basically do some sort of a aggregation of the information and sending only the aggregated informations which lowers the basic bandwidth requirement intermediate bandwidth requirement also lowers the data load at the Cloudant so what we see it is a technology which is a need of the hour and especially with IO T's and other things coming in a big way so with this we will stop today thank you

One Comment

  1. Suruj Uddin said:

    Thanks about your activity and your valuable lecturs.

    June 28, 2019

Leave a Reply

Your email address will not be published. Required fields are marked *