 Hi, this is Hosebni Bhartiya and welcome to T3M, our topic of this month. And the topic of this month is data. And today we have with us, once again, Aditya Madan, director of products at Elixio. It is great to have you on the show. Hi, Satnil. Glad to be back. And the aspect of data that we're going to talk about today is data gravity or egress fee, which makes it very expensive for organizations to move their own data around. Well, the fact is that all major cloud providers allow you to move data into the cloud for free. But if you want to retrieve the data, if you want to migrate or move the data around, then they impose heavy fee. It's a hidden fee. It's fine print. It's known as data egress fee. Aditya, explain to our viewers, what is data egress fee? As you know, a lot of times what happens is storing data in the clouds is cheap. And that's what the cloud providers incentivize enterprises to do, that you get all your data to the cloud. And then you can make use of all the different compute services that the cloud might provide. But anytime the data moves either across regions of the cloud or when it moves out of the cloud, either for access by on-premises data centers or by access by a different cloud, they charge you based on the amount of traffic that moves across the network. And that's what we call egress fees. And this concept is the same regardless of the cloud provider that we talk about. What kind of organizations should worry about data egress fee? Is it a concern for large organizations who are dealing with huge amount of data that they need to move around? Or is it applicable to anyone putting any amount of data in the cloud? So definitely egress fees is not a concept that applies to all enterprises. It really depends on the scale of the enterprise and the complexity of the enterprise. There are some natural reasons for data to be distributed. A lot of times if an organization goes through mergers and acquisitions and there are natural reasons for data being siloed across different locations, then there is a need to worry about egress fees as everyone wants to get insights or drive revenue from the collection of data that they have. So that is one situation. And another situation these days is as people are accelerating their move or their utilization of AI ML, including deep learning, availability of GPUs is a reason for compute to be separated from the data naturally. And that's another reason if you're ramping up your AI ML initiatives, that's another reason to pay attention to egress fees because you might want to compute data away from where it's located or away from where the gravity holds it. We talk about data click, data warehouse or data lake house now. Companies do move data around for various reasons. It could be compliance around geographical reasons. It could be within department. It could be to extract value from the data leveraging right AI ML platforms and technologies. Can you talk about what are organizations currently doing to deal with this data egress fee? Yeah, I think overall if you just look at the complexity of such a platform, whether it's a combination of a lake, lake house or a warehouse, keeping things all in one place and operating all in one place is of course the simplest option out there. So if you put all of your things in one cloud vendor and everything is being done in that one cloud vendor, that is going to be the simplest solution for you. But if the economics does not permit, if there are services that you want to use from another cloud like the availability of GPUs, you will be more spread out. So once this situation occurs, once you become a little more spread out across different locations, a typical approach that we've seen being employed is people would manually copy data across these two environments. And the reason why copying is preferred option is because direct access incurs egress fees as we were talking about before. Data is repeatedly accessed. In the case of model training, for example, you would access the same piece of data, maybe even up to maybe a hundred times. So it's repeatedly accessed and that's why copies is something that's employed. But copies needs you to maintain a team. You would have four or five people to just for this it's error prone and it has its own set of problems. And that's kind of where we come in with our solution to really simplify this process and also eliminate not just the network traffic but also which causes the egressive but also the complexity that goes along with it. When you look at some of these practices, are they kind of leading to some new patterns or trends? The way organizations are looking at their data? Yeah, so these days that the trend that we are seeing with AI ML initiated is kicking off in a lot of organizations and it being especially competitively important for them not just to drive revenue but also just survive as a company is we are seeing the rise of specialized systems to tackle the high throughput and performance requirements of GPUs and such. So in this case, again, from the trend that we're seeing from the cloud providers as well is to have a uniform solution which goes across your data lake which is where all of the data lands but also the consumption of it not just for analytical purposes in the past but also for model training and deep learning and machine learning as well. If I ask you what advice do you have how organizations should approach data to cut down on data egressivity and escape data gravity? And also if you can talk about how is Elixir helping them? One of the advice that we give our customers is that make sure that you are not having to maintain any redundancy in your platform. And by that, one of the things I was touching upon before is even if the same piece of data is being consumed for analytical purposes or it's being consumed for model training purposes, specialized systems are going down that route of having special purpose systems just for one specific purpose. It's something that we've seen not work out repeatedly. That's kind of why we went from specialized data warehouses to what we call data lake houses and the flexibility that we offer with that. There are some very specific situations which demand that but as much as possible, having eliminating redundancy within your platform will go a long way. And especially because with the amount of initiatives that are kicking off, you don't know which direction will scale faster, which specific need within your organization will scale faster. And where Alexio comes into this is one way of eliminating redundancy is having something like us which serves multiple kinds of workloads in your organization. So we propose a solution which sits between different kinds of compute frameworks, whether it's for analytics or machine learning and we're able to serve a single copy or from a single data source, multiple different application types without having to move into a specialized system. Of course, we always talk about technology and culture. Does it also require some cultural changes within organizations when it comes to how we should look at data or technological solutions are enough? I think that the culture aspect is definitely there. So I think how you form your organization and the culture could refer to a lot of different things. One aspect of a culture could mean how you embrace open source and how open the technology stack that you're using. For example, another part of culture could be how do you structure your teams to specialize on different aspects? Like do you have one platform team which serves both AI and analytics? Do you have special, do you structure it in a way that you have specialized teams dedicated for both? So not just the technical side, I think technical side are just the tools that you have available to be more productive. But to your point, there's a lot on the cultural side as well. On the first one that I was talking about with being more open, different things would make sense for an organization at different times. It may make sense for you to use a completely bundled up out of the box service at one point, but you may want to migrate to something which is more open over time as your platform scales. Thank you so much for taking time out today and talk about this topic. And as usual, I would love to chat with you again soon. Thank you. Thank you, thumbnail team here.