Lecture - 42 Resource Management - I

Summary

In this lecture, the focus is on resource management within the cloud-fog-edge paradigm in computing. The discussion highlights the evolution from cloud computing to incorporating fog and edge computing to handle challenges like latency and efficient resource allocation. It introduces the importance of processing data closer to the data source rather than purely relying on cloud solutions. The lecture provides insight into the necessary infrastructure, control mechanisms, data flow, and algorithms that optimize resource management in these systems. It also covers practical examples and use cases, such as healthcare and vehicular communication systems, illustrating the role and challenges in deploying effective resource management strategies.

Highlights

• Introduction to challenges in resource management and service placement within the cloud-fog-edge paradigm. 🔧
• Discussion on the significance of latency and resource constraints in computing environments. ⏳
• The role of fog and edge computing in improving data processing efficiency and reducing cloud dependency. 🌤️
• Insights into hardware, system software, and middleware needed for fog-edge infrastructure. 🖥️
• Various algorithms for discovery, benchmarking, load balancing, and service placement are explored. 🧩

Key Takeaways

• Cloud-fog-edge computing optimizes resource management and reduces latency by processing data closer to its source. 🌟
• Resource management is crucial for balancing workloads and maintaining performance in the fog-edge-cloud paradigm. ⚖️
• Effective algorithms and infrastructure are vital for successful resource planning and execution in distributed systems. 🛠️
• The integration of fog and edge computing enhances the capabilities of traditional cloud setups by addressing real-time data processing needs. 📈
• There are challenges in interoperability, control, and resource allocation that require sophisticated solutions. 🔍

Overview

The lecture begins by addressing the evolving landscape of cloud computing to include fog and edge paradigms, which help in processing data closer to where it's generated. This shift is driven by the necessity to overcome latency and inefficient resource allocation inherent in traditional cloud-only scenarios. The cloud-fog-edge model aims to improve real-time computing and reduce bandwidth usage by handling data at local nodes.

Resource management in these distributed systems is critical. It involves balancing loads, provisioning resources, and ensuring performance across different computing layers. The lecture delves into the mechanisms needed—hardware, software, and middleware—that enable these systems. Considerations include the centralized or distributed nature of control and the hierarchy within these systems.

Practical applications highlight the importance of these paradigms in industries like healthcare and transportation, where timely and reliable data processing is crucial. The challenges of integrating various technologies and ensuring interoperability call for complex, well-designed solutions. This reflects the need for advanced resource management strategies to optimize performance and efficiency in cloud-fog-edge architectures.

Chapters

• 00:00 - 00:30: Introduction and Overview This chapter introduces the course on cloud computing, going beyond just a general overview to delve into more relevant, advanced topics.
• 00:30 - 05:30: Challenges of Resource Management The chapter discusses the challenges of resource management in the context of cloud computing, particularly focusing on the service placement problem in cloud and fog computing environments. It builds upon previous lectures that provided an overview of cloud computing, zooming into specific challenges within resource management.
• 05:30 - 10:00: Fog and Edge Computing Paradigm The chapter discusses the Fog and Edge Computing Paradigm and its significance in achieving improved resource management and performance metrics such as timing and latency. Key concepts addressed include resource management issues and service placement problems, setting the foundation for exploring other related topics.
• 10:00 - 15:00: Resource Allocation and Cost The chapter focuses on the concept of resource allocation and its associated costs, particularly in cloud-based scenarios.
• 15:00 - 20:00: Latency, Reliability, and Interoperability This chapter discusses key concepts such as latency, reliability, and interoperability in the context of health parameter monitoring. It highlights the role of body sensor networks and various sensors in measuring health parameters. These measurements are then uploaded to the cloud for computation, which can return the results to a mobile device, display unit, or other devices. The chapter likely explores the importance of ensuring that these processes are both efficient and reliable, and how different systems can work together (interoperate) effectively.
• 20:00 - 25:00: Fog Layer and Interoperability The chapter discusses the limitations of processing IoT applications directly through the cloud, especially for time-sensitive tasks that require more efficiency and involve large amounts of data. It introduces the concept of fog computing as a promising alternative, highlighting its potential advantages over direct cloud processing.
• 25:00 - 30:00: Resource Management Architecture The chapter titled 'Resource Management Architecture' discusses the paradigm shift in computing, which involves processing large amounts of data closer to the sources rather than relying on the cloud. This approach raises important considerations for resource management within fog environments, highlighting that it's not always a straightforward advantage to increase reliance on cloud computing.
• 30:00 - 35:00: Fog and Edge Infrastructure The chapter discusses the expectations and challenges associated with fog and edge computing infrastructures, especially in relation to IoT devices. It highlights the misconception that these infrastructures will automatically manage IoT devices efficiently without issues. However, in reality, some resources might be over-constrained while others might have spare capacity, necessitating effective resource management.
• 35:00 - 40:00: Resource Management Algorithms The chapter discusses resource management algorithms focusing on scenarios where cloud resources may not be available. It highlights the importance of managing resources efficiently, including cost considerations. Resources management involves not just the allocation of resources but also managing the time necessary to perform tasks effectively, particularly in fog and edge computing layers. The need to manage resources at these lower layers is emphasized.
• 40:00 - 40:30: Conclusion The conclusion chapter focuses on several key elements of resource management, particularly emphasizing resource allocation, workload balancing, and making sure resources are equitably distributed. It highlights the importance of resource provisioning and task scheduling while maintaining appropriate Quality of Service (QoS) levels.

Lecture - 42 Resource Management - I Transcription

• 00:00 - 00:30 [Music] hello so let us continue our discussion ah on this cloud computing course rather we are discussing about little bit more relevant topics or which are little bit above the general overview so what i
• 00:30 - 01:00 believe that you have already seen the other lectures which gives a ah overview of the overall cloud computing thing and today we will be mostly looking at that the challenges in resource management also the problem related to service placement in a cloud ah cloud fog age type of thing this lectures as i mentioned in my previous ah this ah couple of lectures as i mentioned in my previous talk that we will be concentrating more on cloud fog
• 01:00 - 01:30 paradigm rather cloud fog age paradigm and why it is important to realize a better resource management better performance in terms of say timing or latency and type of things so mostly we will be looking at concepts like resource management issues and also we look into the service placement problems rather ah this will be our base for going to other type of ah things like
• 01:30 - 02:00 where we will look in subsequent lectures ah into things like ah migration issues and other type of things so these are keywords so ah just to quick recap what we have seen that only cloud only type of scenario that means you have this underlining iot or sensor devices and above at the other end the cloud so
• 02:00 - 02:30 like you have some of the things like ah health parameter like the probably the type of things we discussed last lecture that you have body sensor networks or different sensors which has this ah different type of health parameters and that it uploads in a cloud it computes the whole thing and ah return the result to the either to the device of the mobile device of the display unit or any other
• 02:30 - 03:00 display unit of the patient or the client right ah so what we have seen that processing iot application directly to the cloud may not be always efficient solution ah because specially for time sensitive things right where it requires more time and not only those things it it have a lot of resources right like lot of data being communicated over the over the this network so a promising alternating what we have seen that making a fog edge
• 03:00 - 03:30 computing thing ah paradigm so these paradigm imposed to process large ah amounts of generated data close to the data sources rather than the cloud so the what we have seen that it is by the definition it is bringing down so one of the consideration of ah this cloud based fog environment is the resource management definitely right like ah so it is not always always a win-win situation like you you go on increasing
• 03:30 - 04:00 your iot devices and then you expect that your age intelligence or the fog infrastructure will take care that may not be true not only that in number of cases what we will see what you what is been seen that some of the resource may be over constrained and some may have still something spare so appropriate managing these resources is also necessity there is a
• 04:00 - 04:30 another point to be noted whenever we want to do any type of management thing which was may not be there in the cloud to any scenario it also brings about ah some costing right so the a fog ah resource managers purses should work and that fellow will also it awaits not only some resources ah it also it takes time to do the things right so nevertheless ah some sort of a resource management at this ah down the layers fog edge type of layers is
• 04:30 - 05:00 necessary that that is ah there right ah and it is a sort of a resource allocation if so this amount of resources are there how it will be allocated ah work load balancing right so it should be somewhat equi distributed as far as possible right ah resource provisioning task scheduling q s ah maintaining appropriate q s level and type of things are coming into ah
• 05:00 - 05:30 play so in other say sense we see there is a whole lot of things coming to play nevertheless irrespective with ah it is means in spite of ah handling this type of scenario this type of paradigm where ah cloud with fog and age may be beneficial that we want to see so our todays talk is mostly hover around ah how this resource management things will be ah what are the different parameters or
• 05:30 - 06:00 what are the difference ah means means characteristics of this resource management problem so what ah we just to continue from the previous slide so fog edge to support for the cloud computing paradigm so the major ah issue is the latency is right like may involve transfer of data for each single sensor to the cloud so whereas this if we have
• 06:00 - 06:30 this paradigm it will reduce that communication ah likely to reduce that means as we have seen that i can aggregate the data in cases if it is possible and then ah transmit some of the things i can response at the fog or edge layer itself like some of the call it may not have to go to that level to things like as example we were discussed earlier like if i want to look at the room temperature and want to make a alarm call that the temperature is above the threshold level that may
• 06:30 - 07:00 not has to take the go up to the cloud level ah and take a call ah that my even my sensor if we have a more computing things that means at the edge itself i can take if the temperature is more than something it ah sets up the alarm and say that it is there sometimes it is aggregation is required like i all the sensors in this room should agreed upon things like it may happen that one of the sensors may be misbehaving and set up a alarm and
• 07:00 - 07:30 whereas if there is aggregated things so that could have been done in a little much higher level not maybe the age those fellows are sensing may be at a at the fog level again i ah reduce the load on the cloud right or sometimes we want to make a aggregate ah guessing of this data and then send on the cloud so the overall traffic from this fog to the cloud reduces right so this is these are definitely latency issue and so fog edge computing made cloud in
• 07:30 - 08:00 overcoming this so it is not a substitute ah rather ah it is a supportive or a cooperative or collaborative system came into play and there are a lot of challenges also come into play right once you collaborate there are issues of interoperability there are timing relationship and between these different processes and things like ah this comes into play right
• 08:00 - 08:30 rather ah what we will see in this subsequent lecture this event this fog layer may help you in attain in making ah this different type of devices interoperate right fog edge layer right like i for example if i am taking say even say come to our example of temperature sensor if we if even if we look at the temperature sensor that may be a variety of there
• 08:30 - 09:00 may be more than one type of ah sensor that means coming from different make and model so what will happen it may end up in sending simple or as ah basic as that they may be sending out these temperature things in a different format for that matter suppose one is sending in centigrade and one is ending in fahrenheit like now ah this making in a uniform conversion could have been done in the ah either in that
• 09:00 - 09:30 edge itself or at the fog layer so in other sense it allows me to interoperate in the things and if my if there are variety of sensors so you may ah can imagine the complexity of the whole ah thing right so that this three technology can work together to grain grant what is improved ah latency ah reliability in some cases like ah and faster responses right so though these things are ah
• 09:30 - 10:00 interrelated ah things are apart from that there are that will help in ah at times in interoperability like to interoperate between the different things and later on we can we will see ah that it also may help in ah some sort of a um what we say information security right now like some of the things like fog etcetera may be in my
• 10:00 - 10:30 control or in my premises so while sending the data i can have ah i aggregate the data so individual data is not thing ah in is not ah pinpointed or even i can strip off this ah say patient information that present ah like patient identity and send that data up in the thing and while coming back i attach it i number it with the things right so starting from that there are other crypto etcetera we can run we are not
• 10:30 - 11:00 going to those complexity nevertheless what we see that it will definitely help in in having latency ah reliability and faster response and may at times this trustworthiness and type of things so ah again we look at the same [Music] picture or same type of things so we have several ah ah devices or iot sensors and actuators like sensing and actuating based on things and intermediate we can have a
• 11:00 - 11:30 fog layer right so there are ah different fog layers and if you can see this fog layers can have different type of resources like in this cases they are having different virtual machines right which takes care of the things so it requires a hardware a hypervisor and the virtual machines once you have you need to have a fog manager ah server manager to manage this if there are more such things you need to have ah a some sort
• 11:30 - 12:00 of a a fog sort of a fog data center or a fog network which also may require a some sort of a management of the things right there are other things what we will see that there are there are challenges of mobility right like as as we our age devices or this iot things can be mobile so this fog thing should take care of the things in other sense if it is under one paradigm of the wong fog device right we are considering the fog devices
• 12:00 - 12:30 to be stationary then once it cross and go to the another fog device ah area so that there there should be a sinking of the things that same sensor is sending data or if there is a data aggregation going on i may not do i may not want to lose this ah data type of things right so those things are there there is underlining communication network definitely which puts it to the cloud so there is a request goes on rather there should be other way around as it is actuating some of the some ah it is request and this
• 12:30 - 13:00 other arrow is the request response type of things so it sends some request and get a response and actuate on the different type of autonomous mobile sensors and type of things right so what we see that fog environment model typically is composed of three layers a cloud client layer edge a fog layer and a cloud layer right fog layer is to accomplish requirement of the resource ah resources of the client and if there is no limited ah
• 13:00 - 13:30 no or limited availability of the resources in the fork layer then the request to be passed to the ah cloud layer right so there may be two scenario where this ah edge is pushing to the fog fog is pushing to the cloud one is that it do not have the capability of computing that type of things right that ah particular ah analysis or that ah computing model or it is getting constrained on the resources once it is getting constrained in your resources basically in case of effort it can
• 13:30 - 14:00 delegate to the other ah what is corroborative fog devices right or it can push it to the cloud another thing is that it do not have the ah way to compute this right like ah if i want to say if if we looking at the say all temperature sensors in all over this building all rooms and then want to take a call that or in this whole our institute and then want to take a call that which
• 14:00 - 14:30 which are the things which are ah maintaining and which are not maintaining that individual fog devices and aggregately how many percentage of the this air conditioning systems are working faithfully or in a proper way so that can be so this ah global thing ah can be done at a ah cloud level right so the major functionality so ah what we see if we have this type of
• 14:30 - 15:00 fog edge layer so major functionality is that the there should be a fog fog server manager which employs ah all available processors to the client right and there are definitely vms operate on the fog data server process theme then delivers to the result to the fog manager and fog servers content for ah server manager virtual machine to manage this using a server virtualization technique etcetera so in other sense the type of virtualization techniques and
• 15:00 - 15:30 computing things that comes down below that the ah fog layer from the ah cloud infrastructure so that may be a ah overall ah managing this fork layer right so this ah type of things may make lot of sense in ah application specific domains right it will change over time like if i if i if we look at ah say say something like regular ad hoc
• 15:30 - 16:00 network type of things right where where we have underlining this cards and vehicles which are having their own ah sensing units and not only that they are having ah that on board units or obu's and it communicates with either another car right so if i if i look at these are the edges right with sensors at the down so it can either communicate with the
• 16:00 - 16:30 things what we say v two v communication vehicle to vehicle or it ah communicates with the roadside infrastructure so v two i infrastructure right this roadside infrastructure is connected to a back end ah cloud data centers center so for all type of things suppose a ah revocation of one license etcetera or something is there certificate is there then it is basically taken a call at a much higher level and done like that something is misbehaving or something is
• 16:30 - 17:00 happened ah means something is some group of car is misbehaving like slowing down not expected so it may have to be taken a call that whether there is a in whether it is a there is a call in there there is a accident in front and type of things which may have to be correlated with other informations which is known to the cloud right so that type of things are there so what is happening now this ah this infrastructure may maintain different resources where ah different virtual
• 17:00 - 17:30 machines etcetera ah there and it has a separate ah things which acts as a fog layer for all those things so which push it to the cloud cloud also can definitely have other means resource management things right so it can be different so this is one scenario if we look at the scenario like health care systems then we have say if we have a body area network so it captures ah things transmit to a nearest hop or a it may be a ah a local server or even a mobile
• 17:30 - 18:00 device which acts as a ah fog and in turn it push it to the back end health cloud and type of things right it can be that way also so in case of a say regular network it can be too much mobile right it is ah make and breaker there where in case of a other scenarios like either health or like temperature measurement or temperature humidity and etcetera measurement across
• 18:00 - 18:30 the rooms those are more or less static sensors they are not moving around ah now and then so the trend as we can see is to centralize some of the computing resources available in large computing server things that is why we are having this fog as we have seen dedicated micro data centers and take internet nodes such as routers gateways switches are augmented for the computing facilities and ah another set of definition i kept it
• 18:30 - 19:00 little ah in italics that because you may be by a time bombard with so many definition but nevertheless the philosophy is still remains same bringing down the computing from the cloud to down the to the as much as to the client end or the where the sensor end so that the computing is more facilitated right so if we look the overall ah resource management paradigm so this is
• 19:00 - 19:30 the thing which we referring that reference is given below ah a nice very nice survey paper which takes care of these ah it deals with lot of things will be ah primarily looking at that ah stuff so ah if we look at that ah resource management in fog and edge computing rather for cloud fog and edge computing so we like to divide it into three major components right one is ah architecture wise how they are
• 19:30 - 20:00 ah infrastructure wise and algorithm wise which is dealt in this particular reference paper you can look at the whole document ah very ah nicely written survey paper and pretty decent one ah end of 2019 so ah and if we look at everything like architecture so architectures also have different components like segment like data flow where as we are discussing that aggregation of the data
• 20:00 - 20:30 or sharing and offloading of the data right so like aggregating the data if it is make sense sharing information for other computing purpose or things and offloading to ah other devices or of loading to cloud and type of things control another aspect is the control right which looks as though whether the control is a centralized control even even if you look at the fog layer itself whether it is a centralized
• 20:30 - 21:00 control or distributed or it can be hierarchical control right i can have different layers of hierarchical control so this now we are having this fog as things in a much bigger perspective when we look at the infrastructure so infrastructure whether we are looking at hardware infrastructure or system sort software or middleware right so different aspects are being studied or lot of research going on that how overall things works right so
• 21:00 - 21:30 ah on the other side other side given that we have a infrastructure and architecture in place what are the different type of algorithms which will come into play like ah things like discovering ah the resources or benchmarking the resources or how things will work load balancing issues or load balances ah algorithms and there are challenges of placement right given a given a job
• 21:30 - 22:00 where do i place right the job even if you can see that even those who have gone through distributed network distributed computing type of param paradigm ah they are also what we have seen that this placement of the things is ah are challenges rather all these components are somewhat present in those type of scenarios so ah if we look at the resource
• 22:00 - 22:30 management approaches one is the architecture as we are discussing used for ah resource management in ah cloud fog edge computing ah based on data flow or classified data flow control and tenancy right but where it will be there where it will be where it will be deciding right so tenancy infrastructure the again infrastructure composed of as we have seen hardware and software to manage the thing network storage resources for
• 22:30 - 23:00 application ah and ah this like ah overall resources how it can be managed right and on the other side we have the algorithms so underlining algorithms to facilitate ah cloud fog edge computing paradigms so these are the things which are which are very much needed for this type of scenarios right
• 23:00 - 23:30 so again coming back to the figure if we want to have resource management so these are the major component and where ah it need to be we need to looked into now we will let us see ah one component ah other so architecture if we look at the architecture as we have seen ah it is data flow right control and tenancy so it is based on the direction of the movement of workloads like when we talk about the data flow ah in the computing
• 23:30 - 24:00 overall computing ecosystems right so workload can be transferred from the user to the devices ah to the age nodes or alternative the cloud server to the edge nodes and type of things so how this data is flowing ah based on your ah application requirement ah also dictates how the resource need to be managed right like if i say my my data is from the user end to the age and fog and
• 24:00 - 24:30 they are it likely to be computed and come back then i have to make the resource management that way if i want that my the if we the requirement is the data has to travel to the cloud and something has to be done ah at the cloud and come back then i have to take care of that so called intermediate ah this computing network ah resource management also into play vis-a-vis how this data being transferred etcetera right the other part is the control so
• 24:30 - 25:00 based on how the resources are controlled in the computing overall computing is for a single controller or a centralized algorithm may be used for managing a number of edge nodes or it may be a distributed things it all depends that how much your overall paradigm is right and the thirdly what we see when you look at the architecture issues based on the support provided for the hosting multiple entities in the ecosystem ah either a single application or multiple
• 25:00 - 25:30 activation could be hosted on the edge so that means if we look at this architecture so one is that one looks into the ah one aspect looks into the overall data flow mechanisms other is the control ah mechanisms like how overall resources can be controlled whether it is in a centralized way or in a distributed way suppose i have a one or a set of fog devices more or less within the things then i can do in a
• 25:30 - 26:00 centralized way or if i have say a for network itself then i may have a distributed control into the things right there may be issues that based on the applications like ah like say regular network type of scenario where i may re i may want a distributed computing because there is a that the vehicle is moving over a large ah geographical space and things like that and need to be looked into whereas in a other type of scenario
• 26:00 - 26:30 where temperature control etcetera which are more or less static things we i may look for a distribution so it also depends on that and then tenancy means which algorithms which will be running and where they will be reciting and where it will be running is a important thing right a single applicant or multiple application ah hosted on edge nodes or which things will reside on the things so that is also important that where where what will be the tenancy
• 26:30 - 27:00 if we look at the infrastructure as we have seen the other aspect to the infrastructure so there is a things called hardware ah system software and ah middleware right so when you look at the hardware is the [Music] ah things like if we look at fog edge type of paradigm so it exploits the small form factor devices such as network gateway wi-fi router set top
• 27:00 - 27:30 boxes small home servers age isp servers and so on and so forth and if we say regular things like car and vehicles even drone as a computing servers for resource efficiency ah things like that right so that means if i having this ah other resources in place like if we look at wi-fi router or in a vehicular communication car or in a drone capturing images or other information so drone so if we can add on computational
• 27:30 - 28:00 module or computational resources it has a additional computational or unused computational resources i can utilize we can utilize that for ah for this fog edge computing right typically cloud have that ah sufficient resources to work on right and the other aspect is the system software runs directly on the fog edge hardware resources such as cpu memory and other
• 28:00 - 28:30 devices right so once the hardware is there based on that my system software runs right so it has there should be a underlining system software which manage the things it manages resources distribute them with the fog edge like operating system virtualization software and type of things right so those are the things these are systems of stress which are not ah per se actually the application specific software right so these are more running the running the devices and type of things and then we have a middleware
• 28:30 - 29:00 which runs on the operating system and provides complementary services that are not supported by the system software right like middleware coordinates distributed ah computing nodes and performs ah deployment of vm and containers to each etcetera so these are the middleware which are not typically support the jobs which are not typically type supported by the system software but it supports this type of things as we are
• 29:00 - 29:30 seeing that ah deployment of vm or container in the thing so those are those are the things which are middleware right which cut across ah maybe ah more than one ah resources right so if you look at the infrastructure so these are typical component of infrastructure and finally ah in this paradigm if we look at that algorithms so dis discovery benchmarking load balancing and placement so these are the four major components when you look at so what we
• 29:30 - 30:00 say infrastructure is in place ah the architecture based on the architecture and now we have the algorithms to run over and above the things right so discovery is identifying the easy source so workloads from the cloud or from the user devices can be deployed on them right so this algorithm ah i should have underlining according that which resource is free and which are the edge devices where it can run so bench marking is capturing the performance like ah it may be cpu it may
• 30:00 - 30:30 be storage device networking and type of things of a computing this so it is very much needed because ah this actually dictates that how overall ah this overall performance over the of the system will be there and load balancing as age data centers are deployed across the network edge issue of distributing task using ah say efficient load balancing algorithm has gain gaining significant ah
• 30:30 - 31:00 important like so it may deploy different optimization technique load balancing techniques and different type of searching mechanisms and type of things so it is important to have the load balancing otherwise ah sometimes it may be skewed into the things right it if it is if it is going to the seating the same type of devices so this the algorithm need to take care
• 31:00 - 31:30 correct and finally the challenges of ah address placement the placement of services and applications right addresses the issue of ah of placing ah incoming computing task on suitable resources so how i can place this resources considering available resources in fog edge layer and environment [Music] if there is a environmental change right
• 31:30 - 32:00 so it can be dynamically condition dynamic condition aware techniques or iterative techniques so i can we can have different type of things so what we see that this ah four components is is the algorithmic things so now again if we come back to this ah big picture right where this resource management with respect to fog age rather cloud fog edge type of ah paradigm is there so one aspect is looking at that what sort of architecture we are having which has
• 32:00 - 32:30 data flow control and tenancy type of ah issues to be handled other ah part is that the infrastructure hardware system software middleware how things are there and ah and then given this scenario we need to have this appropriate algorithms ah to run on those things so that the applications are there so our basic bottom line is that this our different sensors which are being there i want to take a call based on this data right whether it
• 32:30 - 33:00 can be done at the age intelligence or at the fog layer or at the cloud based on the things if i want to use this more efficiently this fog age type of paradigm along with the cloud i need to have a proper resource management right we will continue our discussion ah on this topic or more precisely some of the areas in my sub in our subsequent lectures to see that how overall this cloud fog
• 33:00 - 33:30 age paradigm ah helps us in achieving ah better performance ah in the overall what we say this ecosystem thank you