Ahead in the Cloud: Harnessing Open Source for Innovation with Rafee Tarafdar

Chad Watt: Welcome to Ahead In The Cloud, where business leaders share what they've learned on their cloud journey. I'm Chad Watt, an Infosys Knowledge Institute researcher and writer. I'm here today with Rafee Tarafdar. Rafee is the Chief Technology Officer of Infosys and a leader in thinking at the intersection of technology and business. He is joining us today to talk about open source software for large enterprises. I'm delighted to have you here today, sir.

Rafee: Thank you so much, Chad. Thanks for having me on this podcast.

Chad Watt: Just to set some parameters, when we talk about open source software, we're broadly talking about programs where source code is visible and can be modified, shared, and used for any purposes. And we are recording this at the start of summer 2023, and everything in tech continues to focus around generative AI, and Infosys itself just launched our own suite of AI-first services called Topaz. In a world where everyone's focused on AI, how is it that open source is relevant in this kind of climate of AI-first technology programs and product launches?

Rafee: I think great question, and in my view, open source is still equally relevant even with all the evolution and development we are seeing in the AI space. If you look at all the AI models and products, the way we see it is coming out in three different forms. One is closed-access models. These are models like GPT from OpenAI, Bard from Google, which are not directly available for organizations to download and use, but they are available through APIs that you can consume and these are being monetized in some cases. Second is open access models. These are the likes of BLOOM, CodeGen and others that you can actually download it. You have the weights available, you know the data sets that they are using. Now we have seen even StarCoderBase that has come up recently for code. So a lot of these models are being open source, which just like any other open source software, I should be able to download and use it and also fine-tune it for my specific purposes and I can contribute back.
The third is, in terms of these specialized AI products and specialized models, a lot of these are anyways getting producted, but I think going forward we will see a lot more of these open access models, we will see a lot more of open source software and platforms that will come up, which will make it easier for enterprises to adopt in addition to the closed access model and products.

Chad Watt: That's very interesting, thinking about those three sorts of models and the closed access model, of course, is the first one we all saw last fall in terms of the generative AI chat and the image generators. If I can ask you which one of those three you think is going to produce the most useful tools for business right now, what would your wager be?

Rafee: I think if you look at it, the first two are essentially the core ingredients, which are the models. Today if you look at it, a lot more powerful models are there in the closed access model space, but we are now seeing equally good come up in open access model space as well. So I think both of them will flourish and both of them will coexist.
A lot of specialty models, specialized models or industry-specific models, for example, specialized for healthcare, specialized for financial industries, or for indexes and all that, I think these will all become protected or these will remain closed access, whereas a lot more general purpose one will become available as open access. The third anyways is the AI products. I think these anyways will be built using a combination of the closed access or open access models. While some of them will be open sourced, I think a lot of them will be monetized through some monthly fees and so on and so forth.

Chad Watt: Right. It's very interesting thinking about the ways in which people can deliver value and derive value from kind of open source and collaborative models. Now, open source has actually been around for quite a while. It's older than cloud computing itself now, but open source and cloud are tightly linked in the modern technology applications. Do you see a similar sort of linkage happening between open source and AI down the road?

Rafee: No, absolutely. I think we'll see both of them fairly well intertwined, and I think a lot more innovation will happen today when these are two intertwined. Because while the tech majors are anyways doing a lot of innovation and they're sharing a lot of research for wider good, but the power of the community is that they can innovate at scale and hence I think we'll see the value come out in both the cases.
And this is what see we saw even with the cloud era, right? Because in cloud, if you look at it, a lot of innovation on containers, a lot of innovation in the middlewares, a lot of innovation with the kind of web servers, app servers, all of databases, and even for the data warehouses and data platforms. All of these came from the open source community and then they become mainstream. And we also then have foundations like Cloud Native Computing Foundation. We are enabling this in a very structured way on how these open source technologies can help accelerate the value that we get from cloud.

Chad Watt: Very good, very good. What are the most critical ways open source and cloud work together for business, especially large enterprises today? What are one or two of the best ways you see open source being put to work in cloud?

Rafee: Yeah, see, I think I'll take an example from Infosys itself, right? And if you look at Infosys, we today run some of the world's largest tax platforms in India for the government. And the volumes, if you look at it, are very, very significant. It has about a hundred million members. And with any tax system, most of the filings are typically done during the last one or two days.
Now, the entire platform in itself has been built largely on open source technology. It has been built on using all the cloud native concepts, architecture and design patterns, and it is made fully scalable, fully resilient, and there is significant amount of observability at scale and all. So you can imagine when you are building and running a population-scale platform fully on open source and on commodity hardware, that's the kind of scale that today the open source software is able to support and deliver.
And also when I look at it on the enterprise or business side, today, most of the businesses significantly leverage the container technology. They leverage a container-based databases, container-based solutions, and then a lot of orchestration is done on open source, a lot of databases run on open source. So a lot of businesses today are also running their mission-critical workloads on open source at global scale. So I think today, open source itself is one of the most important ingredients on the cloud in addition to the cloud native services that each of the hyperscalers provide.

Chad Watt: That's very fascinating, especially when you think about the kind of legacy of open source. The heritage of open source, if you will, is it's the mom and shop, it's the freeware, the hobbyist, the bulletin boards, the people who kind of, it's mom and pop software. But that's not the case today. This is powering big business and large enterprise. So what are the three or so most common misconceptions about open source software?

Rafee: I think I would say more than misconceptions, there are a few factors that a lot of business leaders are concerned about when they have to use open source for mission-critical applications. One is the resiliency of these at scale. So for example, let's take a database. If I have to run my entire platform on a database, then do I have support to fail over? Do I have the support to do data application at scale? Do I have the support to make sure that it can handle the volumes? Do I get a production support in case there is a production issue which needs to be resolved within an hour or within a few hours? Do I get quality support from the freemium vendors who are supporting these open source projects? So that's one level of considerations that a lot of them think about.
The second is also in terms of the licensing policies and what you are supposed to do or not do with open source. And today, if you look at it, there are a number of different licensing models that are available. So a lot of organizations have to be very careful in making sure that they're using the right licensing type and the software and then whatever solutions or platforms that they are building on top of the open source tool, they just need to ensure that they are adhering to the policies that are defined by it.
The third is security, because one of the things that happens is today we are in a world where a lot of issues come up on an ongoing basis, and whenever any issue is identified, we need rapid action in order to fix this. Like a few years back, there was a vulnerability identified in Log4j. Now what it then meant is every software that was using Log4j had to be patched or fixed, and this means you need a significant action in order to very quickly address these issues.
So I think these are the three, in my view, top concerns as well as considerations that most of the enterprise leaders would think about.

Chad Watt: Security, licensing, resiliency. And you feel confident that open source, there are options out there for any sort of business that could put these to work?

Rafee: Yeah, absolutely, if you make the right open source choice. So I wouldn't say that every single open source software can provide all of it, but we have actually run a lot of mission-critical platforms at population scale as well as at enterprise scale using open source and we are very confident that when the right software and stack is chosen, it can deliver the outcomes that the business expects.

Chad Watt: How long have you been thinking about and working with open source, Rafee?

Rafee: As an individual, I've been working with open source for almost about two decades or so. While initially, the number of choices were very less and now the choices have increased significantly. Now if you look at Infosys as an organization, while we have been working on open source for a very long time and we've been using open source in a very strategic way, over the last few months we have now decided to be more intentional and strategic about it.
So one of the things that we have done now is that we have set up an open source program office at Infosys. And the idea of setting up an open source program office was to make sure that as a company, we are able to establish a framework based on which we can not only nurture and drive usage of open source within the company, but we also become very active contributors to the entire open source world.
And as part of it, there are five focus areas that we focus on. One is to make sure that there is a process and a framework and a policy that is defined on how to go about using open source, how to go about contributing to open source, and how to go about open sourcing projects in a way that it doesn't violate any of the IP of our clients' attention, ensures that we are following all the IP and legal regulations and policies defined.
The second is to essentially create a community of open source contributors, open source engineers, and today we have about 75,000 odd engineers who are very conversant with open source, who are using it and who are also contributing back to the open source community either by providing some defect fixes or by providing some patches or contributing new modules.
Third is we also want to become active contributors where either we can contribute to an existing project, we can contribute a new module to one of the open source programs, as well as open source some of the existing Infosys projects.
Then at a strategic level, we also decided that we will partner with two of the big open source foundations. One is Cloud Native Computing Foundation, and with the Linux Foundation for Networking, so we've become platinum members in both the foundations. And the goal for us is that how do we take the voice of our customers, voice of enterprise users to the foundation and to the open source community so that the open source community and foundation are aware of the kind of challenges that enterprises face and we are able to address them very well. And also as we build roadmaps, we are able to take some of those needs and requirements into the roadmap.
And the last is to create an ecosystem where we can have an ecosystem of open source partners, open source contributors, and look at bringing all these innovations in a holistic manner. So a lot of these, I think, in a strategic way, we have now been doing for the last few months, and there is a lot more that we will do over the coming years.

Chad Watt: Is there something that a typical enterprise can do better with open source than they have been doing with proprietary? Can you put your finger on something that's a good example of where open source has become this choice over a proprietary system?

Rafee: Today if you look at it, I think, containers is a good place where open source containers are fairly strong and most of the cloud providers use it. If you look at database technology, I think a lot of what comes from open source is fairly robust. Most of the businesses use either open source or cloud provider-managed databases, which are based on open source and otherwise.
If you look at a lot of tooling that is used in DevSecOps, a lot of tooling that is used in MLOps, and a lot of tooling that is used in experience development like Angular, React and all this stuff. So most of these frameworks that are used in application development are also open source based. If you today look at a lot of web servers and others, they're largely based on open source. So I think you will find that one is a lot of tooling at infrastructure level, a lot of tooling at a database level and at an integration level.
So for example, Kafka today is fairly standard in any enterprise wherever they have any need to do any synchronous or asynchronous communication and all that stuff. Same thing you know will find in experience, the frameworks, utilities, SDKs have become fairly standard. So I think these are the areas where we find that these are filled. When it comes to verticalized applications, when it comes to business functionality, when it comes to business process, this is where I think the open source has still not made that level of impact. So a lot of times for core business functionality and core business services feature functions, you would still rely on either the SaaS platforms or the build it on past platforms or use the CORTs products that today run on cloud as such.

Chad Watt: Thank you again for your time today, Rafee. This podcast is part of our collaboration with MIT Tech Review in partnership with Infosys Cobalt. Visit our content hub on technologyreview.com to learn more about how businesses across the globe are moving from cloud chaos to cloud clarity. Be sure to follow Ahead In The Cloud wherever you get your podcasts. You can find more details in our show notes and transcripts at emphasis.com/iki. That's in our podcast section. Thanks to our producers Catherine Burdette, Christine Calhoun, and Yulia De Bari. Dode Bigley is our audio technician, and I'm Chad Watt with the Infosys Knowledge Institute signing off. Until next time, keep learning and keep sharing.