The tail end of 2023 has seen its fair share of cutting-edge enterprise technology announcements, leaving the industry in awe of the sheer pace of innovation.

Amidst attention-grabbing unveilings, such as Graviton4 and single zone storage class for Amazon S3, another update slipped under the radar.

“What was not mentioned was that S3 has something called Access Grants, which do very low-level fine-grain access control at an S3 level,” said Sanjeev Mohan (pictured), principal at SanjMo. “Not even mentioned, nobody’s mentioned it because there are so many announcements. If you start looking into the details, Amazon has really innovated across … everybody wants to hear generative AI, but it’s across the stack.”

Mohan spoke with theCUBE industry analyst Dave Vellante at the “Supercloud 5: The Battle for AI Supremacy” event, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed the cloud landscape, where each announcement — whether on vector databases, Zero-ETL or language translation — unfolds as a piece of a complex and ever-evolving puzzle.

Zero-ETL expands, and so does MongoDB’s Atlas Vector Search

Zero-ETL, a concept introduced last year, aims to simplify data pipelines. The acronym stands for “extract, transform and load,” and it’s the process of combining, cleaning and normalizing data from different sources to prepare it for analytics, artificial intelligence and machine learning workloads.

In the context of Amazon Web Services Inc.’s data platform, the Zero-ETL principle has seen monumental strides, particularly in the integration of DynamoDB and OpenSearch. This integration extends to vector search, a pivotal move that aligns with industry trends toward seamless, configuration-based data aggregation, according to Mohan.

“MongoDB has introduced vector embeddings inside the JSON document,” he said. “DynamoDB does similar stuff, but instead of putting vector database or vector embedding inside DynamoDB, [AWS] have actually extended Zero-ETL to OpenSearch. OpenSearch, as you know, is an open-source version of Elasticsearch, which already has inverted search. Now it has vector, so you can even do semantic search.”

Additionally, a pathway has been opened for the strategic use of Zero-ETL in Aurora, MySQL and Postgres. It becomes evident that for customers managing complex data pipelines, Zero-ETL serves as a game-changer, ushering in a new era of efficiency and agility in data management, Mohan added.

“Now with Zero-ETL, it becomes easier because now you can go into Redshift, just configuration and it’s done for you,” he explained. “All of a sudden, now your data is available in Redshift for aggregations.”

Another key development is MongoDB Inc.’s Vector Search. MongoDB’s approach of incorporating vector embeddings inside JSON documents outshines its counterparts, according to Mohan. Importantly, MongoDB has decided to offer Vector Search as a free feature embedded within the core platform. This move positions MongoDB as a formidable player in the market, especially when compared to other database vendors selling the feature as a separate capability.

“MongoDB has some very distinct advantages because their foundational database is really strong,” Mohan explained. “For example, they introduced earlier the separation of search nodes. So, you can do separation of concern, you don’t overload — you can scale as well. The vector search can sit on the search node. They’re getting a lot of these advantages for free.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of the “Supercloud 5: The Battle for AI Supremacy” event:

Photo: SiliconANGLE