Authentication, storage, real-time push, database API – 4 back-end modules, traditional development takes weeks. Today, you only need 30 minutes.
Full stack development engineer at an AI startup company. Last week, the boss said, “You will be responsible for the backend of the AI conversation application that will be launched next week” – authentication, database interface, file storage, and real-time push. Each module takes several weeks to build, and the market will not wait for you.
Xiao Zhang opened Huawei Cloud RDS Supabase, set up authentication, database API, and object storage in 30 minutes, and spent the rest of the time polishing the AI interaction experience.Not because of working overtime, but because of using the right tools.
Huawei Cloud RDS Supabase AI application is officially launched for public testing.Based on open source Supabase + Huawei Cloud RDS PostgreSQL full hosting capabilities, it provides developers withOut-of-the-box BaaS (Backend as a Service) service——No need for additional development and deployment of back-end application services, effectively reducing labor costs and resource expenses.Development and deployment cycle shortened from months to daysto achieve efficient and agile development.
👉 Pass it on to the developers around you. It’s time to change the back-end building thing.
Today, with the explosion of AI applications, back-end infrastructure construction has become the first hurdle faced by developers – not model selection, not algorithm tuning, butHow to quickly build a reliable backend?
Supabase is an open source backend as a service (BaaS) platform, built with PostgreSQL as the base.Taking into account the powerful capabilities of relational databases and the flexibility and freedom of the open source ecosystemis the first choice for open source to get rid of the lock-in of commercial BaaS. It deeply adapts multiple mature open source frameworks to PostgreSQL and provides one-stop core capabilities such as data API, user authentication, object storage, and real-time communication. Developers can quickly build high-performance and scalable applications without building back-end infrastructure from scratch.
Traditional application architectures often have multiple layers of coupling – front-end, back-end services, and databases are tightly bound, and each component needs to be independently developed, deployed, and maintained. Supabase breaks this dilemma with integrated services:
🔹 Backend services
Self-built API layer, tight coupling between front and back ends → Data API is automatically generated, decoupled and speeded up
🔹 User authentication
Manually integrating OAuth/JWT took weeks → Multiple authentication methods are available out of the box for quick access
🔹 File storage
Self-built file system + CDN, heavy operation and maintenance burden → S3 compatible object storage, plug and play
🔹 Real-time push
Polling or self-built WebSocket, high complexity → CDC real-time subscription, zero-configuration access
🔹 Data control
Relying on third-party platforms, data is locked → Data is completely autonomous and controllable
🔹 Development cycle：Monthly level → Heaven level
🔹 Cost investment：high → significantly reduced
🚀 Develop and deliver
Traditional: 3 back-end development × several months cycle
Supabase: 1 front-end development × several days = greatly shortened delivery cycle
In the traditional model, the front-end and the database are coupled layer by layer, and the development and deployment cycle often takes several months. Supabase replaces cumbersome infrastructure construction with integrated service capabilities. From concept verification to product launch, it completes delivery in a shorter cycle and quickly seizes market opportunities.
🔌 Function modules are plug-and-play
Traditional: weeks for OAuth integration, weeks for file storage, days for WebSocket
Supabase: Just a few lines of code each and get it done quickly
Each functional module is independently enabled and combined on demand – authentication, storage, real-time push, and can be accessed with just a few lines of code. There is no need to build complex backends for weeks, significantly reducing the cost of trial and error, so that innovative exploration is no longer hindered by infrastructure.
🔒 Data is completely autonomous and controllable
It takes a long time for commercial BaaS users to migrate data and there is a risk of format loss.
Supabase: your database, your data, export at any time
Built on open source PostgreSQL, data is stored in your own database, saying goodbye to the data locking of commercial BaaS platforms. Row-level security (RLS) ensures multi-tenant data isolation and each user can only access their own data.
|
scene |
core competencies |
Time to get started |
|
🔮 AI conversation application |
Auth + Data API + pgvector vector retrieval |
about 30 minutes |
|
🤝 Real-time collaboration tools |
Realtime zero configuration access + broadcast message |
about 30 minutes |
|
🏢 Enterprise internal management platform |
RLS row-level security, multi-tenant data isolation |
about 1 hour |
|
🚀 MVP quick verification |
From idea to runnable demo, one-stop solution |
several days |
How can such a powerful open source BaaS obtain enterprise-level protection?Huawei Cloud RDS gives the answer.
Building a self-built Supabase requires building a K8s/Docker operation and maintenance system, configuring high-availability primary and backup systems, and handling security compliance. The operation and maintenance cost is no less than the development itself. Huawei Cloud RDS Supabase adds enterprise-level enhancements to open source capabilities——Let Supabase change from “can use” to “dare to use”。
|
Dimensions |
Self-built Supabase |
Huawei Cloud RDS Supabase |
|
Deployment method |
To build K8s/Docker by yourself, you need to operate and maintain 9 major components |
Create a console with one click and complete it in 3 steps |
|
High availability |
You need to build your own master and backup and switch manually. |
RDS automatically switches between active and standby, RPO is nearly zero |
|
Security Compliance |
Configure SSL/VPC/security group yourself |
Security group + DNAT + SSL one-click management |
|
Data disaster recovery |
Self-built backup strategy, recovery takes a long time |
Full backup + additional backup + PITR point-in-time recovery |
|
life cycle management |
Command line operation and maintenance, high operating threshold |
Visual control of the entire process, what you see is what you get |
|
Access method |
Configure internal and external networks and domain names by yourself |
Dashboard, API SDK, AI Agent MCP (API Key), ready to use out of the box |
|
Operation and maintenance monitoring |
Community support, self-diagnosis of problems |
OPS monitoring alarm, HA self-healing |
RDS Supabase AI application providesOne-stop fully managed servicecovering the complete life cycle from application creation to operation and maintenance management:
- Full life cycle management: Create, start, stop, restart, delete, visual control of the entire process, what you see is what you get
- Enterprise-level security controls: SSL certificate, security group, DNAT gateway, password management, multi-layer protection to ensure data security
- Multi-channel easy access:Dashboard visual operation, internal and external network access, API Key management, flexible adaptation to various usage scenarios
- Supabase core capabilities are available out of the box：Data API, Auth authentication, Storage storage, Realtime real-time, four major capabilities can be accessed with a few lines of code
- Huawei Cloud RDS PostgreSQL enterprise-level guarantee: Highly available active and backup, high performance, automatic backup, security compliance
From a technical implementation perspective, RDS Supabase adopts a three-layer decoupled deployment architecture, with clear responsibilities and loosely coupled collaboration at each layer:
- Application access layer: Provides four access methods: OpenAPI, SDK, WebConsole, and MCP, covering all scenarios of management, development, operation and maintenance, and AI Agent
- RDS service layer: 9 major Supabase core components (Kong gateway, GoTrue authentication, PostgREST data API, Realtime, Storage, Imgproxy image processing, Studio visualization, Supavisor connection pool, Logflare log analysis) + Huawei Cloud OBS object storage and cloud disk
- RDS business data layer: PostgreSQL active and standby architecture, supporting RLS row-level security, Auth Schema, and pgvector vector retrieval
Huawei Cloud RDS Supabase is based on the open source Supabase + Huawei Cloud RDS PostgreSQL full hosting capabilities, providing full life cycle management from application creation to operation and maintenance management. Developers do not need to pay attention to the underlying infrastructure and can complete all operations on the RDS console.
From creation to deletion, full-process console visualization operations:
- Create with one click:Basic configuration→Hosted instance→Network configuration, complete application creation in three steps
- Flexible management and control: Start, stop, restart, respond to business changes at any time
- safe release: Automatically recycle all associated resources when deleting an app
Data security is the lifeline of application launch. RDS Supabase provides multi-layer security control capabilities:
- SSL certificate management: Reset SSL certificate with one click to ensure transmission link encryption
- Security group management: Refined network access control
- DNAT gateway binding: Flexible configuration of internal and external network access policies
- Password management: The application login password and the managed instance password are managed independently without interfering with each other.
RDS Supabase provides a variety of application interaction methods to adapt to different usage scenarios:
- Dashboard visual operation: Intuitively manage all functions such as database, users, and storage through the web console
- Intranet access: Suitable for internal service calls in the same VPC, with low latency and high bandwidth, ensuring that data does not leave the domain.
- External network access: Suitable for development, debugging and remote management, efficient access anytime and anywhere
- API Key authentication: Supports new version (recommended) and Legacy keys, flexibly adapting to different client integration scenarios
- AI Agent docking:Operate and manage the database in an AI-friendly way through the MCP protocol
Development is delivery – data is the blood of the application. Supabase takes data from “needs development” to “ready to use out of the box”.
Supabase is based on the PostgREST framework and automatically maps database tables to RESTful APIs. You can create a table to get complete add, delete, modify and query endpoints. There is no need to write DAO layer code, just focus on business logic.
Get started quickly
// JavaScript——查询数据（过滤+排序+分页）
const { data } = await supabase
.from('users').select('id, name, email')
.eq('status', 'active').order('created_at', { ascending: false }).limit(10)
# Python——查询数据
result = supabase.table("users").select("id, name, email") \
.eq("status", "active").execute()For complete CRUD examples and more usage, please see Supabase official documentation
Supabase Storage provides an object storage service that is compatible with the S3 standard, supports high-performance upload and download of large files, and combines Auth’s RLS policy to achieve precise file access permission control – who can see, who can upload, and who can delete, all with one policy.
👆 Storage management interface – Bucket management, file upload and download
Development is delivery – interaction is the soul of the application. Supabase takes authentication and real-time communication from “an integration problem” to “a few lines of code”.
Supabase Auth provides complete user registration, login, and session management capabilities, supports multiple authentication methods such as email/password, mobile phone number, OAuth third-party login (GitHub, Google, Apple, etc.), SAML SSO single sign-on, and implements fine-grained permission control based on JWT.
Each authenticated user is bound to PostgreSQL’s row-level security policy (RLS) through a role (Role) to achieve multi-user data security isolation – for example, in resource management applications, you can set “users can only view resources created by themselves.”
Interface display
👆 Auth management interface – user management, policy configuration, login log
Get started quickly
// 用户登录
const { data } = await supabase.auth.signInWithPassword({
email: 'user@example.com', password: 'secure-password'
})
// OAuth第三方登录（以GitHub为例）
const { data } = await supabase.auth.signInWithOAuth({ provider: 'github' })Realtime is Supabase’s most differentiated technical capability—the CDC mechanism based on PostgreSQL logical replication. Any data written can be captured in real time and pushed to all online clients in seconds.Say goodbye to polling, changes are instant. Whether it’s instant messaging, multi-person collaborative editing or real-time data dashboards, you can access it with just a few lines of code.
Get started quickly
// 实时订阅——监听表的新增记录
supabase.channel('messages-channel')
.on('postgres_changes', { event: 'INSERT', schema: 'public', table: 'messages' },
(payload) => console.log('收到新消息:', payload.new))
.subscribe()
# Python——实时订阅数据变更
def on_insert(payload):
print(f"收到新消息: {payload['new']}")
supabase.channel("messages-channel").on_postgres_changes(
"INSERT", schema="public", table="messages", callback=on_insert
).subscribe()Why is it called “AI application”? Because Supabase is more than just the backend—it’s the infrastructure for AI applications.
pgvector vector retrieval
RDS Supabase has built-in PostgreSQL pgvector extension, which can complete structured data storage and vector retrieval in the same database without deploying an additional vector database——One database handles all data requirements of RAG applications。
-- 创建带向量字段的表（需先启用vector扩展）
CREATE TABLE documents (
id UUID PRIMARY KEY, content TEXT, embedding VECTOR(1536)
);
-- 向量相似度查询——一条SQL完成语义检索
SELECT id, content, 1 - (embedding <=> '[0.01, 0.02, ...]') AS similarity
FROM documents ORDER BY embedding <=> '[0.01, 0.02, ...]' LIMIT 10;
# Python + LangChain 集成
from langchain_community.vectorstores import SupabaseVectorStore
vector_store = SupabaseVectorStore.from_documents(
docs, embeddings, client=supabase, table_name="documents")
results = vector_store.similarity_search("如何使用Supabase构建AI应用？", k=5)MCP Agent access – let AI Agent directly operate your database
RDS Supabase supports MCP (Model Context Protocol) access method. AI Agent can directly operate the database, manage users, read and write storage through MCP protocol——Let your AI Agent have “database hands”。
- 🤖 AI Agent directly calls Data API through MCP, without the need for additional development interface layer
- 🔐 Agent operations are subject to RLS policies to ensure data security isolation
- ⚡ Combined with pgvector, Agent can realize semantic retrieval-driven intelligent decision-making
MCP docking effect demonstration
Through the MCP protocol, AI Agents (such as Claude, ChatGPT, etc.) can use natural language to drive database operations to achieve an interactive experience of “say a word and check a batch of data”. The following are real cases:
🗣️ Developer: Query the current business library list overview
🤖 AI Agent: Call the tool supabase.list_tables to scan the current database table structure
✅ Return results(Core business table):
|
business table |
use |
|
products |
Product master list |
|
product_pricings |
product price list |
|
product_media |
Product multimedia materials |
There is no need to write a line of SQL or open a database client. AI Agent can get the full picture of the business database in just one sentence——This is the true value of MCP’s “natural language direct drive database”。
💡 How to experience it? Configure the MCP Server address and API Key of RDS Supabase in the AI client that supports MCP, so that the AI Agent can directly read and write your database. Detailed configuration tutorials will be released in subsequent articles, so stay tuned.
Development is delivery – management and control are the guarantee for launch. RDS Supabase Dashboard changes operation and maintenance from “command line operation” to “visual click”.
RDS Supabase provides a visual dashboard that aggregates core capabilities such as database management, user authentication, file storage, and real-time communication into the same interface. Developers can complete all daily operations without switching tools.
👆 Overview of the core functions of Dashboard – one-screen control of five major modules
|
module |
Function |
Typical operations |
|
Table Editor |
Visual table management |
Create tables, edit data, and set relationships |
|
Database |
Database management |
View table structure and manage extension plug-ins (including pgvector) |
|
Auth |
Users and Authentication |
User management, policy configuration, login log |
|
Storage |
File storage |
Bucket management, file upload and download |
|
SQL Editor |
SQL query |
Execute SQL online and save query templates |
