Powering the Digital Experiences You Trust

Distributed SQL Summit - Asia 2025 ©2025 All Right Reserved
Yugabyte: Powering the Digital Experiences You Trust Vish Phaneendra Vice President, Asia Paciﬁc and Japan

2 Shoyu Traditional & Reliable Miso Rich & Scalable Tonkotsu Flexible & Powerful

3 The Ramen Philosophy of Databases Many flavors, each excellent alone. But what if you could have one bowl with the best of everything?

4 Digital Life Runs on Data Every seamless experience, from tapping your Suica card to receiving real-time updates, is powered by a reliable and fast database working behind the scenes.

5 Yesterday's monolithic databases, like the first Shinkansen, were revolutionary. But they can't power today's always-on, globally connected world.

6 YugabyteDB is the modern Shinkansen for your data: cloud-native, distributed, and built for zero downtime. It brings resilience to your most critical workloads without sacrificing performance.

Customers & Partners around the World Trust YugabyteDB to Run their Business

Asia Key Customers 8 8 Financial Services Telco & e-Commerce Automotive, Maritime

9 Our Triple Pledge Reliability "Like a Shinkansen — always on time." Scalability "Like 7-Eleven — everywhere, same quality." Flexibility "Like Japanese cuisine — adapts to taste."

10 Hello from the Year 2027! A journey through the AI application architecture of the (not too distant) future.

Preview: Lessons from the Year 2027 11 For example: • All your apps will need updating! • LLM models will keep evolving! • Devs/ops need to do more with less. • Wait, is RAG in or out? MCP vs A2A? • Scale apps, but in seconds not months • No manual performance tuning • More options for users to switch apps Key principle: build to last design for change AI will accelerate things beyond our imagination – exponential growth

Lessons from the Year 2027: The impact on software and operations teams “Agile? LOL. We shipped 350 apps this year.” 12

Can conﬁrm! Explosive AI growth in 2027 13 Source: https://www.grandviewresearch.com/industry-analysis/ai-apps-market-report#

AI app modernization put a *huge* strain on 2027 software teams • Expectation: build and iterate AI apps, quickly ◦ Building with familiar Postgres saves resources • Massive amounts of re-architecting to be done ◦ Your architecture and tech stack requires flexibility • Success can be viral, will it scale? ◦ What worked in prototyping and small usage may not scale 14

“IDC predicts that in 2026, over one-third of organizations will be stuck in the experimental, point-solution phase of AI experimentation, requiring a shift of focus to enterprise use cases to deliver ROI.” - IDC, 2025 15 Some companies were left behind last year…

Application Lifecycle (waaaaaay back in 2025) Birth of a cloud-native app Innovation Getting ready for growth Dealing with infrastructure outages Planned and unplanned scaling 16

Application Lifecycle (in 2027) GenAI app Birth of a cloud-native app Innovation Getting ready for growth Dealing with infrastructure outages Planned and unplanned scaling 17

How we do it in 2027: Steps for building a modern GenAI app 18 Existing App + DB 01 Prepare Knowledge Base Collect and preprocess data for retrieval 02 Create Indexes Organize data into searchable indexes 03 Set Up MCP Server Deploy server to interface with the knowledge base 04 Advertise Server Capabilities Ensure server functionalities are known 05 Configure MCP Client Integrate server with AI environment 06 Create A2A Application Enable agents to communicate and collaborate +

Focus on the core: data and indexes 19 01 Prepare Knowledge Base Collect and preprocess data for retrieval 02 Create Indexes Organize data into searchable indexes 03 Set Up MCP Server Deploy server to interface with the knowledge base 04 Advertise Server Capabilities Ensure server functionalities are known Existing App + DB + 05 Configure MCP Client Integrate server with AI environment 06 Create A2A Application Enable agents to communicate and collaborate

AI-powered apps - not built to last, but designed for change! 20 Load data PgVector does not support generating embeddings) \COPY public.ybarticles FROM 'File.csv' Create tables with vector columns CREATE TABLE images( id INTEGER NOT NULL, title TEXT, image_vector vector(1536)); Create vector indexes with different distance functions CREATE INDEX ON images USING ybhnsw (image_vector vector_l2_ops); Search for vectors closest to an input SELECT id, title FROM images ORDER BY image_vector <=> '[-0.01,-0.02,0.01...]'::vector LIMIT 10; HNSW Index Different distance functions Vector Search

Vector Index Flexible RAG app architectures thrive in 2027 21 Additional documents Query Response LLM Encode Embedding model Index Similarity search Prompt Similar documents Query RAG Encode 1 2 3 4 5 6 7 Existing App Document Chunks App Data

Choose your infrastructure wisely to accelerate change 22 Disadvantages of Standalone Vector Database Advantages of YugabyteDB as a Vector Database • API-based, prone to desync. • May offer proprietary indexing algorithms. • Lack full database features ACID compliance, row-level security). • Faster query times, but with potential network latency. • More cost-effective for smaller, proof-of-concept POC) projects. • Requires additional effort for data synchronization and management. • Built-in PostgreSQL data sync. • Supports multiple indexing algorithms (e.g., IVFFlat, HNSW • PostgreSQLʼs full ACID backup, and security. • Higher accuracy and often outperforms in QPS, especially with HNSW indexing • More economical at scale, leverages existing PostgreSQL infrastructure • Seamless integration with PostgreSQL data and familiar tools

Oh, sorry, in 2027 we say “ROG”, not RAG Retrieval-Optimized Generation ROG • Leans on the database to reduce (expensive) LLM utilization • Reduces latency, which 2027 users expect • Focused on your business data first and foremost • Requires multi-modal approach 23

PostgreSQL continues to grow in 2027 24 PostgreSQL is used by 49% of developers and is the most popular database for the second year in a row. Source: Stack Overflow 2025

25 However, software teams need capabilities beyond PostgreSQL Cloud-Native Enterprise-Grade AI-Ready

PostgreSQL Compatibility and Cloud Native Architecture are Critical Can use PostgreSQL client drivers and psql shell Parse PG syntax - but execution is different Syntax Supports some advanced PG features - but they will work differently Feature Exactly like Postgres. Port over all existing apps, PG developers instantly at home. Runtime Wire How much Postgres compatibility? How cloud native (distributed) is the architecture? Low Medium High Can beneﬁt from Postgres innovation (like pg_vector for genAI, QoS for multi-tenancy, etc) Can innovate on distributed, cloud native architecture (like zero downtime, global apps, fast auto-scaling, connection scaling, etc.) Can innovate on both dimensions PG Innovation Threshold Cloud DBMS Innovation Threshold Cannot deliver high data durability, availability, scale, best in class DR Delivers data durability and some vertical scale. Weak HA, horizontally scale, DR High data durability, availability, scalability, DR, multi-region 26

PostgreSQL Compatibility and Cloud Native Architecture are Critical Can use PostgreSQL client drivers and psql shell Parse PG syntax - but execution is different Syntax Supports some advanced PG features - but they will work differently Feature Exactly like Postgres. Port over all existing apps, PG developers instantly at home. Runtime Wire How much Postgres compatibility? How cloud native (distributed) is the architecture? Cannot deliver high data durability, availability, scale, best in class DR Low Delivers data durability and some vertical scale. Weak HA, horizontally scale, DR Medium High data durability, availability, scalability, DR, multi-region High Cloud DBMS Innovation Threshold PG Innovation Threshold 27

YugabyteDB makes PostgreSQL enterprise-ready 28 Postgres YugabyteDB High availability and DR Global database, xCluster Online upgrades and downgrades Yes Blue-green deployments Yes Security and SecOps Yes Scalable CDC Yes Connection pooling and scaling Built in Observability and Perf Tuning Agentic App Multi tenancy Yes (more coming) YugabyteDB is bringing capabilities available in commercial RDBMS to the PG world.

Lessons from the Year 2027: Why do we need scale and Ultra Resilience? 29

Six elements of Ultra Resilience for future apps In-region resilience Multi-region BCDR Zero-downtime operations Data protection Peak and freak events Grey failures 30

Immediate business value for modern apps Reduce Cloud Risk High Availability Reduce Support Costs Increase Customer Sat Increase Operational Efﬁciency Reduce Technical Debt Reduce Upgrade Stress Automate Backups 31

First data, now LLM…new laws and regulations id geo 1 US id geo 4 UK id geo 2 IND 3 IND • For regulatory and compliance purposes • Pin rows of a table or indexes to speciﬁc regions • Strong consistency • Low read and write latency in-region 32

Applying lessons from the Year 2027: How do I use this information? “Here’s what you should do RIGHT NOW!” 33

34 LlamaIndex Langchain Vertex AI Bedrock OpenAI LocalAI OLlama Cursor Checkout Tutorials at: https://docs.yugabyte.com/preview/tutorials/ai/ Greater ﬂexibility with expanded AI Ecosystem

35 It’s easy: Building AI apps with the Yugabyte MCP Server • Bridge the gap between your AI application (and its LLM) and your YugabyteDB data • Intuitive data exploration, without needing to write a single line of SQL • Provides a secure, extensible foundation for LLM-enhanced applications

Agentic AI for Ops: YugabyteDB Performance Advisor 36 • A single load chart to provide an overview of system activity • Anomaly correlation helps users to understand how detected issues relate to overall performance • Query-focused analysis allows users to see how specific queries contribute to database load • Interact with Performance Advisor using natural language chat

Ready now (not in 2027)! YugabyteDB extends pgvector indexing Native Vector Search powered by pgvector • In-database vector similarity search • Query vector and relational data • PostgreSQL compatibility • Fully ACID compliant • Fully open-source • Support for multiple distance functions, indexing methods Scalable & Extensible • Built natively for YugabyteDB, Vector index automatically sharded across nodes for horizontal scale and high throughput • Pluggable indexing design enables integration with new algorithms and libraries as the vector search ecosystem evolves • Integration with other Postgres extensions to enhance value Resilience & Global Distribution • Inherits all of YugabyteDBʼs fault tolerance, high availability, and geo-distribution for low-latency vector search across regions 37

AI-ready, multi-modal Distributed PostgreSQL for business-critical applications. Run modern business-critical applications, including GenAI apps, with built-in resilience, seamless scalability, and flexible geo-distribution. 38

Thank you

Powering the Digital Experiences You Trust

Powering the Digital Experiences You Trust

More Decks by YugabyteDB Japan

Other Decks in Technology

Featured

Transcript