AWS re:Invent 2021: Four key takeaways for ASEAN firms

Local Zones and Outposts

AWS has been extending its footprint in key geographies not only by establishing new cloud regions but also through Local Zones, a type of public cloud infrastructure deployment model first announced in 2019 that places compute, storage, databases and other services closer to its customers.

Each Local Zone lets enterprises connect back to the rest of their workloads in the parent region, addressing the needs of applications with low-latency requirements.

In Southeast Asia, the Philippines will be among 21 markets that will be getting a Local Zone starting in 2022. Dutt said while there are many customers in the Philippines that are using the Singapore cloud region, there are cases where “low single-digit millisecond latencies” are required.

“The typical use cases that we see are in media and entertainment, where you have lots of media files that need to be transcoded on the fly as well as real time gaming,” he said.

To bring cloud computing capabilities even closer to the edge, AWS is also making AWS Outposts available in smaller 1U and 2U form factors (as opposed to a full 42U rack) to cater to branch offices, factories, retail stores, health clinics, hospitals and other sites with space constraints.

Dutt said in Southeast Asia, for example, AWS has customers such as retailers and supermarkets that require compute locally but don’t have room for a mini datacentre in their stores. Vietnam recently joined the list of seven ASEAN markets where Outposts is available.

Graviton3

AWS introduced the third-generation Graviton processor, offering up to 25% better compute performance, up to two times higher floating-point performance, and up to two times faster cryptographic workload performance compared to AWS Graviton2 processors.

“While some people believe that compute is no longer a differentiator, our customers in the last two to three years have been telling us that they really believe that the compute offering from AWS is a differentiator for a variety of reasons,” Dutt said.

One reason is the cost-performance advantage of Graviton processors. Dutt said cost in the cloud world is not calculated in binary terms where one service is deemed to be more expensive than another.

“It's an entire spectrum of factors and one of them is price-to-performance ratio," Dutt said, noting that with Graviton2, which features a socket over chip architecture and more compute cores per nanometre, customers can expect a 40% boost in performance at 20% lower cost.

With Graviton3, which builds on the success of Graviton2, Dutt said customers can expect additional compute performance, particularly for machine learning workloads.

Amazon Lex Automated Chatbot Designer

Building for the future where machine learning is expected to be embedded in applications rather than called out as a standalone development, AWS has introduced Amazon Lex Automated Chatbot Designer. Currently in preview, the tool lets developers automatically design chatbots from conversation transcripts.

“Let’s say your agents have had lots of chats with your customers. If you feed the transcripts to the chatbot designer, it will automatically look at those intents without necessarily having human intervention,” Dutt said.

In insurance, for example, Dutt said it is common for people to file claims, cancel automatic payments, or renew policies with an agent – and each one of those conversations has a transcript history. “Customers that are building chatbots can reduce the time to do so from days to a few hours simply because it’s completely automatic,” he added.

IoT TwinMaker

The use of digital twins to simulate business processes and create models of real-world systems and environments has been growing in the region.

Dutt said while some companies like Singapore’s Zimplistic, the maker of the Rotimatic bread-making machine, are using internet of things (IoT) technology to track how their devices are doing, using digital twins will take those capabilities to the next level.

To that end, AWS announced IoT TwinMaker to make it easier for developers to build digital twins to optimise building operations, increase production output, and improve equipment performance.

“If you have a smart mixer at a customer home, apart from getting in data, what IoT TwinMaker can do is to create a 3D model of it and get sensor data on the motor’s rotations per minute, power and temperature.

“The idea is that if the customer ever has a problem, you can be more proactive because the maintenance scenarios and scheduling can be taken care of automatically,” Dutt said.