Emerging Trends in Cloud Computing

Yash Shah (00:02.702)

Okay. We had to receive notifications saying we are live, but let's hope that we are. And my phone's on airplane mode. So I won't receive the notification, but let's hope we are live and get started. we have one viewer. we are live. Hello and welcome to Momentum Officers. My name is Yash and I'm joined by my co -founders, Harsh, Jay and Kaushik to discuss topic of the week.

emerging trends in cloud computing. Our goal is to provide you with actionable insights and practical strategies that you can apply to your own products. Throughout the session, we encourage you to engage with us by asking questions and sharing your thoughts. This is a fantastic opportunity to learn from each other and give you insights that can help drive your product initiatives forward. So let's get started. Jay, how are you doing today?

also trying to figure out what to do next. How can I improvise this meeting? But today we're going to talk about emerging trends in cloud computing. And so for all the people who tuned in, no points for guessing who will be taking the lead on this one. So Harsh, let's start with the basic one. What is cloud computing? OK.

OK, so let's start with the example. So it's like we can say Google or Microsoft, right? They have built their company, So they have also their server, database, network, right? Currently, they have their physical servers, right? They have their own data warehouse with their warehouse where they are data centers, where they store the data, servers, CPU, all these things, right? Now, like the users, just like us, yeah,

founders, want to build a SaaS product, right? So they have good idea, they have tactic to write the code, right? When the things are going where they are thinking to deploy it, right? When they want to go over the internet, right? So it's not every time possible to there by the physical CPU device, data, this thing, they create the one small data center in their office, right? So where their where the resources they are required, so that resource

Yash Shah (02:26.732)

database, networking, the server side, when you can users can buy these things over the internet, right or from any cloud service provider. So these cloud computing where you don't have to maintain or build or buy any physical device, you just have to because it pays a go model, right? Where you just pay the pay the requirements which you have, that's it.

you don't have to maintain the physical device which you need for your own product. Yeah. So this is a small definition you can say for the cloud computing. So correct me if I'm wrong. Does it mean that instead of purchasing compute power, I'm basically renting compute power as a service that someone else has purchased? So as an example,

Amazon Web Services or Google Cloud Platform or Microsoft Azure or Digital Ocean. So all of those have purchased and built a compute power which over the internet I'm able to purchase as a service instead of actually paying for it. Is that a fair way to look at it? Yeah, yeah, perfect, perfect. And the thing is key, if you have a physical device that it can't be said it can't be said between organizations, right? But if we have AWS cloud, they have

created one server and now it can be shareable between the organization. But you have specific, we can say the boundary of this, but the actual physical servers that can be shareable between the organization. Yeah, right. So, yeah. So, I could, I have been, you know, I've been active and lived in lot of time. like what I would see is there a discussion on multi cloud strategies to optimize.

cost and like also ensuring the data redundancy. Could you brief us little about this thing on like how having multi cloud strategies implemented in one's own product would be helpful? Sure, sure. So before going to multi cloud, let's first understand the public cloud and the private cloud. Okay, so let's consider if we have in -house servers or CPU, right?

Yash Shah (04:49.344)

or we have all the infrastructure in -house. Then it's called as a private cloud. That can be available between organizations. All the data are stored in our network. It can be accessible over the internet. But when we have the public cloud, so public cloud means the actual data can be accessible over the internet. So let me give you an example. So we can say AWS, Azure, Google GCP. All these things are public cloud.

so that anyone can buy their cloud requirements from the cloud service provider over the internet. We can say, such like Oracle, yeah, we can say IBM. They have their own cloud, but it is a private. They are only associated with there. So data cannot be accessible over the internet. If you want to access the data, they have their own network. So they have two things, private cloud and the public cloud.

such like the why the concept of multi cloud is coming right currently so we can say any such like if we want to deploy a SaaS product right so we have different services such like the we need CPU to utilize the servers right we need the data Metabase server to store the data we need some we can say storage where we store the medias right we need some email service there they are helpful to send them right

These are the old services which we need for any SaaS products. Now, we can say there are AWS Azure AGC. This old service provider can provide this service. But we can consider which is the best service provider for my use case. So it's like we can say if I want to use MySQL database.

So what is the best cost efficiency service provider that will give me the low cost, right? And if you want to use the NoSQL database, that can be possible. The different service providers have the good service support things. We can say the price will be good, right? So in multi -cloud, we can say we are using the different public cloud service providers, such like the AWS ECB for my different risk cases.

Yash Shah (07:15.064)

So like, let me give you example. In AWS, we have one service S3, okay, which is useful to, we can say, deploy the static website. Okay, so we can say from my front -end application, I can use my AWS services. Okay, but if we can say on the backend side, my service, if my backend is very strong, right, and if my application enterprise, then I need some, we can say container service.

So at that time, Google had their own Kubernetes services, which is very, very, we can say, the scalability, it's very optimized. So for the back end, we can use the GCP. So this multi -cloud, in multi -cloud, we have used a different public, we can say, cloud for my application. This is known as a multi -cloud. So is it, and so I just repeat what you said, and correct me if I'm misinterpreting it. Would it be fair to say that?

As an example, if I'm planning my wedding, I have two ways to go about it. One is I go to an event management company and then they'll arrange for the decorators, the caterers, the performers, the entertainment, everything. But the second way to go about it is that I would go to one company which is known for really good decorations and then another company which is known for catering and then a third company.

which is able to bring in good performance artists or whatever right from a wedding use case. So first would be sort of a single cloud which is where I go to one company and then they arrange everything for me and I have to use whoever they bring on board. In the second case, I am basically going to whoever is known best for whatever service they are offering. Would that be a fair way to... Yeah, yeah, yeah, sorry, sorry. Yeah, the example is correct way. Yeah.

So there are two clouds, the multi cloud and the hybrid cloud too. So in your example, can say in my home, I have my own background, on my garden, where I can do my wedding. So we can say I have my own space. I don't need any hotel or banquet to do my wedding. But I hire someone, some agency to do decoration for my catering. So we can say

Yash Shah (09:39.232)

some of things have in -house and I buy some services from the public environment. So this is known as a hybrid. So in public we can say all the services, all the things have the public. In multi cloud we can say we use different services from public service provider. Where in hybrid we can say I have some in -house things which is a private cloud.

because of cost optimization or some have better services, I can go to my public service provider to get my services. Right. Well, Harsh, drawing from what the point that you just mentioned. So let's say I'm an early stage, you know, I'm trying to sketch a profile and if that person has a demand, would that, how would, you know, someone like that would go about. So let's say I'm an early stage,

SaaS product which works with AI as one of its core functionalities. So how do I go about choosing my service provider, like from cost, from even security concerns and all those different Performance, right? Yeah, performance. And also the scalability aspect of it, right? If in future if I want to scale, how do I go about? So how do I even go about considering which service provider to choose?

Yeah, so as I told you, each service provider has their own capabilities. We can say some uniqueness, right? Just like all the service providers have the old services, the database servers. We can say the mail service, we can say the elastic radius thing, all these things. Everyone has their own capabilities. But how they will maintain or how the things are, we can say, user -friendly for the development or deploy things.

Set like the we can say, set like the if I have one server, okay. Now, according to my business, know, okay, within a few months, my traffic will be go from 100 users to a million of users. Let's say example, okay. So I want auto scaling things for my server, right? Now, let's go for auto scaling. What is the price range, right?

Yash Shah (12:04.974)

for the set like the AWS or GCP, you can consider this provider, right? So I will compare the things. Next thing is if I want to scale my service, right? So what are the configurations I have to change? Whether it is user friendly, whether they are giving the downtime zero or not, right? So we can say that like, as you said, like you have AI product, right? So which service provider has the better integrations with the AI model set like the GPT, Lama, right?

Like in Amazon, have the Bedrock services which have integrated all the AI models, right? So currently, GCP is on board. But what are the use cases? if you have NLP, right? So it is best to use the AWS microservice to generate the data from the AI, right? So in this way, we can decide which service provider is best for us, whether the AWS or GCP or Azure.

Right. So, the other, Jay because this is, I mean it sort of goes against some of the firm beliefs that I have which is sort of go ahead with one thing and then just put all of your confidence into that one thing. Don't you think that when we go to multi cloud or hybrid cloud, we are just increasing the chances of failure, right? Because

Because now if there are different services that are being procured from let's say one from AWS, GCP, Azure and then even my own server as an example. It just increases the chances of failure to 4X. Because even if one of them goes down, my app cannot run properly. So if I using S3 of AWS and I using let's say some controller service from Azure, if one of them goes down, the app stops working.

Yeah. I think in other way, right? So it's like if you have only one size provider, right? So it's like if you're using AWS for all the services, right? And currently, consider you have selected some US region, right? Where all the services are stored. Now, if this region can be destroyed because of some reason, Then what happens? Tsunami. Tsunami, right? Then all the data, the service, all things are vanished, right?

Yash Shah (14:33.238)

Now just consider if you have divide your application in such a way that you have multi cloud such like the database you are storing in the GCP or you can say the servers are available in AWS service. So actually one of the best advantages of using multi cloud is like the disaster recovery. So like you cannot be relying on one service provider. And in future even if even if such like the AWS have increased their price.

So you have a chance to move from AWS to GCP if you using multi -client because you are not required on only one server. Correct. This is interesting because what just happened in this back end for this, there is a very poor, I mean not poor but not well thought comment that a couple of friends and I use which is basically when I say something and then someone tells me a counterpoint which is so true that

that I am just left speechless. So this is we say how that turns the tables. So what I thought is risky is actually risk mitigation. Which is interesting. So but sorry, sorry, Jay, you were saying something. No, that's completely fine. So, yeah. So the other point is currently we can say the Europe, USA side

the different countries are on regulation, right? Now just consider if I am India based or I am US based, right? But if I want to establish my bridges in Europe, right? Then there I have to follow or I have to some restrictions or GDPR, right? So in this way, we can say this multi cloud can help us so that I can set up my database in Europe, right?

But possible that because of my business is global, I can put my other servers on the US region too. So in this way, we can use the multi cloud too.

Yash Shah (16:39.424)

Yeah, yeah, generally. Interesting, yeah. So just had an open question asked with respect to, so we discussed certain trends with respect to cloud computing, especially a little bit about AI and more about having multi -cloud solutions. And you mentioned a little about the compliance side as well. Would there be any other trends that especially SaaS founders should focus upon right now and utilize it, especially? Yeah, yeah.

Yeah, so currently, think everyone is focusing on the performance, right? Everyone wants my application speed is very fast. want on the disk. I just want to load my page, right? So actually, just consider how this things how the how can get the data faster and faster, right? So currently, the new trend is the you can say the edge computing. Okay, so now just consider you can say

I just want to open a website or like we can say any global because Amazon, right? Amazon. So I just when I opened to Amazon, right? So what do you think the way the actual data is coming said like the product listing, right? The pricing, right? So just consider if the if there is only one we can say the we can say in the US or there is only one city where all the data is stored, right? That can be possible the data.

the users nearby this region can get the data faster. But the users or can say the customers who are very far from this, we can say the data center, they will get the data very late. So this is a very major performance issue.

Yash Shah (18:27.106)

How can we this? By using what we can do is we can divide the big region into small region. And each small region has their own small data center where we can store the data as a casing. So that all the users can have the same performance speed.

And so there are some things also, like how we can sync the data between the small regions, right? So this is the one challenge that can be fixed by the single group. But yeah, so this is a new step for them globally. In over the world, we can give our customers the best service, even if they are in any regions. We can say any city or any state in this area. Yeah. It's so funny because

So as a mechanical engineer and with a person who has keen interest in physics, I know that this data is basically photon packets that travel at speed of light. And it is a testament to our low attention spans. And the fact that we need the data so quickly that speed of light is not enough.

We have to figure out a way to get around the speed of light and physically get closer to the customers or consumers. That even if the data is coming to my phone at that speed, I still find it a little slow. But interesting. So the other thing that I also want to understand is, I've come across this term called serverless computing. when I read it for the first time, I thought, my God, Amazon's going to shut down. So tell me why I'm wrong.

You are wrong, what is serverless computer? Yeah, so before going to serverless, the previous concept, the age computer, let me give an example, okay. Currently, we have the great concept of smart city, So, in smart city, we have traffic signals, right? So, in smart city, the traffic signals are built in such a way that if the traffic is very high,

Yash Shah (20:46.349)

So automatically the time duration of stop and start can be changed. So these things need some real -time data. So like they have good camera or signal things. They capture the real -time data, do some processing, and then give the data, which time can be best used. So this is one of the best examples of edge computing. So in Smart City, they have small, we can say the

small data center in every small region, they will do this calculation and keep the data faster. Right. So yeah, I just, I just came up with this example. Let me give you the example. that, yeah, right, right. Interesting. Now tell us why serverless computing is not going to shut down Amazon. So serverless concepts, if I, it's like the, if I want to use serverless capabilities, right. Then I need a

some service providers like Amazon. cannot like so what is the meaning of serverless? First understand that. So in serverless, it is not like there is no server. Okay. Amazon have their own server, their own database. But think is you don't have to maintain. Just consider I want to I have one SaaS product. Okay.

Now, if I want to use Amazon Web Service, right in the server, not serverless, in using server, right, I have to decide how many CPUs, how many RAMs I need for my server, right? What are the configurations I need for my database too, right? And if the traffic can be high, right? Then I have to increase my server capabilities, right? I have to all these things, I have to learn all this manually, right?

Now just consider the concept where I don't need I don't want to maintain all these things. Okay. I just tell you this is my thing. This is my code base. Now all the traffic's all the configures that can be handled by other services. It's no serverless where you don't have the physique. You don't have to configure the server configuration. Okay. They can but I still need a server. Is it? Yeah. Yes. Yes. So in serverless Amazon have their own server.

Yash Shah (23:12.746)

Okay. Some marketing person, some marketing person in some conference room has been over promising. So imagine if you bought a packet of Lays and it said Poisonless chips. then Poisonless still means that there is poison. But it's still poisonless. It's serverless.

but yeah so yeah so just consider there is a cupboard where you put your clothes right so in we can say in one part we decide I have put my clothes in this box right and the other parts can be used by anyone okay so the box which I have decided I put my clothes here this is not a server I have decided this this box is utilized only by me okay

I put my clothes, I remove my clothes, all these things I do my own self, right? But the other space in which cupboard, so like the according to the users, they are coming to the room, they will put their clothes, they will remove automatically, all these things are done automatically, you don't have to think about it. Yeah, so this is a server, right? So this is a cupboard less cupboard situation. But by the way, in this example, you have

very very accurately describe my cupboard. So I have a box in my cupboard which is where all my stuff lives and then the rest of the cupboard is used by my lovely wife and that's sort of pretty much it. I just manage everything that's in the box and the rest of the things is crazy how accurate it is. You're also married right?

You have a cupboard -less cupboard, OK? A bit cupboard. Interesting. Right, right. Yeah. Harsh, think just drawing back to the question, I mean, the answer that you gave to my other question that I was giving, all these service products, just like how you mentioned AWS, Google Cloud, Azure, all of them give AI and machine learning as a service now. And they do give these, you know,

Yash Shah (25:32.844)

I think they giving these services have transformed the entire industry in past, especially in one and a half, two years significantly, where we are seeing it everywhere. And so let's say if I have a product which doesn't have any of the AI features, but I do want to have some of the other AI features within my product, let's say something as simple as, it just summarizes a particular set of data points that I have.

So how do I integrate with these service providers? What is the process for this? How do I do this? Okay, understood. So firstly, in AI, okay, there are different use cases, okay. And each use case here can be, we can say, can be fixed by some models, okay. So it's like if your product image processing or not, okay. If your product have the natural language processing, they need to say processing or not.

They have some feature or prediction or not, right? So in each particular use cases, there are different services are available. Okay. So it's like the, if I want to predict anything, So it's like, I have the list of inverse data, right? And I want to predict the, can be, you can say the profit of the next quarter, right? If I want to predict this data, so I need some time series data from historical data.

So for this particular use case is there some AI have different model. Okay. And if this model is integrated with the service provider, then we can use by, we can say their APIs or their script code, right? So according to use case, we can have the different models. And currently, I told, Amazon have the Bedrock service, right? Which have the integrated with different models, such like the Lama, GP, not GPT.

Yeah, but Lama, they have, I think for image processing, they also have some model, right? So according to use cases, you can choose which model is best for you. And in this way, you can integrate the AI capabilities in your product. So I decide upon the list of tasks that I have and then decide which task needs some AI capability within them. And then I choose the module that the service providers provide. that's it.

Yash Shah (27:59.532)

Yeah, right, right, right. And other things, there are two things in AI. So first thing, the other AI use case. I have some AI, we can say the AI use case, which I want to integrate in my product. And for that, I can use the GPT or Lama or any models for my use cases, right? Other things which currently trading is the AI in cloud computing. So it's like that we can say,

So just because serverless, right? We have serverless config server. I have serverless model for right. And each request I will, I am storing the data in my log format, right? Okay. Where the time, what are the data they are fetching? What is the URL endpoint? Right now, if I want to, we can say I want to use AI in this data. So it's like, can say which API is called maximum types, right? What is the, what is the maximum response time, which I will get, how to improve my response time.

So in cloud computing, currently, this is a latest thing where you can integrate, configure AI in the cloud computing. So it's like the log format we can say, yeah, we can say in server two, like I want to check how many IPs are, if there are any doors are taken, right? So automatically, will fetch the IP, right? From which IP I will get the more traffic, right?

So in this way, we can use the AI cloud computing. So there are two different things. We can integrate AI for my use cases by configuring the AI models. That can be provided by a sales provider or not. Other things, you can use the AI for your cloud computing. for two analyses to predict the, we can say, predictions in this way.

Interesting. Yeah. Got it. Interesting. So I think that brings us to the end of the conversation today. I think it was a meaningful conversation where at least the three of us were enlightened on. mean, Jay, it really teaches us how market. Yeah, indeed.

Yash Shah (30:21.644)

You can also, you can also, if you your mind to it, you can also write zero cost proposals, like cost -utilized proposals. Yeah. It will still have some cost attached to it, but yeah, but interesting, this is meaningful. So thank you everyone for tuning in and thank you Harsh for taking our questions around cloud computing. I understand that these were very naive and early stage questions, but I hope all the people who joined in were able to.

get some value out of this as well. for all the people who join in, do check us out. We go live every Wednesday. And we'll see you next time. Bye. Bye, guys. Bye. Bye.

‍