How to level up the cloud transformation game

The new era of cloud transformation will entail adoption of multi and hybrid-cloud strategies, deploying AI/ML models, and a continued focus on security and compliance, say experts at TechSparks Mumbai 2023.

How to level up the cloud transformation game

Friday April 14, 2023,

6 min Read

Businesses are no longer just using one cloud provider, as they have dynamic needs. This has led them to adopt a multi-cloud strategy to leverage the unique strengths of different providers. 

The purpose is to drive digital agility and transformation by inventing new ways to accomplish business objectives using the best-of-breed services from different cloud providers.

In a session titled ‘From now to wow: How to level up cloud transformation game’ at TechSparks Mumbai 2023, panellists Gurpiar Sibia, Customer Engineering Leader, Digital Natives Business, Google Cloud India; Kunal Gupta, Vice President of Technology, Magicpin; and Yashoraj Tyagi, CTO and CBO, CASHe; shared insights on how organisations are approaching digital transformation with multi-cloud strategies.

Multi and hybrid cloud for digital businesses 

The panel started with how digital businesses are looking at the speed of innovation, the agility to adapt in changing environments, and the flexibility to make choices. 

“Multi cloud offers businesses the right use case for workloads, a good business continuity plan, and the best return on investments (ROI) across different cloud service providers. It also helps that they are not tied to a single provider and hence get the best of services from public cloud and on-premise infrastructure,” Sibia said.

From a hybrid cloud point of view, businesses that have data residency and privacy concerns and want security-focused operations where the data and apps are on the public cloud and offer a mix of both worlds, this offers a perfect balance. “For regulated sectors, businesses require data residency compliances to be followed, where hybrids can become a good use case,” Sibia added. 

The adoption of multi cloud has also seen a great push, post the pandemic, he observed during the discussion. “Before the pandemic, businesses would debate if they should move to the cloud. Today that thought has changed to ‘which cloud or how many should I deploy’,” he said. 

Cloud bridging the gap and optimising costs

Gupta, of Magicpin, which facilitates local shopping by connecting retailers in the neighbourhood to online resources, explained that cloud technology helped the company bridge the gap for 250,000 merchants in categories like fashion, food, pharma, and groceries in 50+ cities.

Aspects like gamification, building a hyperlocal inventory or catalogue of local merchants by integrating various databases, building seamless payments and delivery systems, and offering insights to merchants about what their customers are looking for/what their competitors are doing have played a key role in driving engagement and commerce for merchants on the platform and their customers. 

“Our technology has been the key to enable these services. From compute to container orchestration frameworks, using AI/ML frameworks for building use cases, working with distributed databases, API and mobile-first architecture, polyglot persistence paradigm, mix of SQL, etc., have been running on the cloud. We have used a mix of technologies, enabled by cloud, to power Magicpin,” Gupta said. 

While execution has accelerated, cloud infrastructure has also shortened the time taken to go from an idea, to a prototype to a product/business. “From building solutions on traditional data centres that took enormous time to even spawn a server and then get the database configured on it, that is now enabled through a click, thanks to cloud,” Gupta said. 

In terms of cloud innovations at Magicpin, he said their ‘magic search’ has more than 900 million stock keeping units (SKUs) that are searchable at a hyper-local level thanks to cloud infrastructure. “Over time, our cloud costs have also reduced by 10% by choosing a healthy mix of working with open-source technologies and services that cloud providers already offer.”

The proliferation of cloud technology has also helped cut latency. Gupta said businesses lose customers if there are any delays during a transaction. “Cloud technologies have definitely helped to solve scale and availability challenges in a faster way, without the need for too much human intervention.”

Building meaningful customer experiences and secured multi-cloud infra

Tyagi, of Cashe, a NBFC fintech platform in credit, wealth and insurtech, shared that being a cloud-first business, a multi-cloud approach has helped build a great customer experience and execute millions of transactions in a day.

“From an availability and resiliency standpoint, it becomes crucial. A multi-cloud strategy gives the flexibility to expand compute strength during increased demand and cuts down on latency issues, compliance, and data residency requirements,” he said. 

From an AI/ML perspective, when in the business of lending to underbanked segments, alternate data sets that are typically not available to formal lending institutions have helped Cashe push the envelope for financial inclusion, by offering credit to those who do not have access to it through formal sources.

“We are able to do this through building large ML models that require scale across cloud platforms. In a multi-cloud scenario, different cloud infrastructures have various strengths that include building with ML models, good container orchestration, or strong ML deployment and fast inferences. This helps use the strengths of each cloud provider to their advantage,” he added. 

Another challenge that cloud technology has helped Cashe is extracting behavioural insights from large alternate data sets that could co-relate to future financial behaviour with respect to credit default.

“This required running large-scale models and took us two to three years to get it right. Model validation is extremely difficult where we have needed deep compute resources, a huge amount of memory intensive jobs to be running on our cloud services, and ML Ops pipelines to make sure that model validation happens regularly,” Tyagi shared. 

He added that it was also important to always keep background processes churning data sets regularly to make sure that the models running were relevant to solutions being built. “That’s where most of our cloud computing resources were used,” he said. 

Future innovations in multi and hybrid-cloud infrastructure 

In terms of innovations in the multi or hybrid-cloud environments, Sibia said containers, modernisation of infrastructure, and data warehouses are important factors to consider.

“Containers are built to be agnostic and they really help to use multi-cloud infra to the fullest. AI/ML not just for data modelling, but to also automate operations across clouds that can offer higher speed and larger scale, zero trust security across multi-cloud are some key trends for 2023,” Sibia said. 

When it comes to the complexity of operating on cloud, he said having the right strategy and partners was essential to help businesses, to avoid taking the same problems from an on-premise to a single cloud to a multi-cloud environment.

“In such scenarios, manageability becomes a nightmare. At Google, we have a history of being multi cloud and open-source friendly. We gave the world Kubernetes and the zero-trust security value. Our products are solving problems across workloads, multiple clouds, containers, and tooling to enable overall transformation for digital businesses with consistent processes and scalable software that can be deployed and run across clouds,” Sibia said.