Built on Nile
A Postgres database that enables all database operations per customer, including auto-scaling, branching, backups, and insights, with full customer isolation. Place customers in any region on serverless or dedicated compute, with the cost and operational efficiency of a single, scalable database
One Postgres database
Unlimited virtual tenant databases
Avoid operational nightmares, frustrating developer experiences and high costs of managing one database per tenant
Eliminate noisy neighbor problems and leaky data isolation when using one database for all your tenants
Enjoy the isolation of db per tenant model with the cost efficiency and developer experience of one db for all the tenants
Customer-specific vector embeddings at 10x lower cost
Use open-source pgvector extension to build multi-tenant RAG applications. Store your vector embeddings and your tenant data in one database
Cost optimized
Built on object storage and shared compute to be 10x lower cost
Limitless embeddings
Scale to billions of vector embeddings across thousands of tenants
Low latency and performance
Deploy embeddings closer to customer and LLMs for latency and compliance needs with one database
1
CREATE TABLE wiki_documents(
2
tenant_id uuid,
3
id integer,
4
embedding vector(3)
5
);
6
7
INSERT INTO wiki_documents(
8
tenant_id,
9
id,
10
embedding
11
) VALUES (
12
'0191c7b8-c62b-7574-b15c-35b1b6fc06fc',
13
1,
14
[18, 24, 43]
15
);
16
17
SELECT embedding <-> '[34, 09, 42]'
18
AS distance
19
FROM wiki_documents;
Secure isolation for customer's data and embeddings
Tenant virtualization
Built-in tenant virtualization isolates data access across tenants. Restrict access to a specific virtual tenant database from application.
Hassle free
No more struggle with Postgres row level security or with complex application-level authorization logic.
Share data
Securely share data across tenants using shared tables
Autoscale to millions of tenants and billions of embeddings
Scale up instantly when tenants receive bursty AI workloads. Scale to zero with no cold start time.
Pay for exact resources utilized by queries to achieve 10x cost efficiency
Limitless connections as you grow your AI use case
Place tenants on serverless or provisioned compute - globally
Use serverless for most of your customers to save cost on your AI workloads
USER IN USA
USER IN ASIA
Tenant
A
Tenant
B
Serverless
Tenant
C
Provisioned compute
Tenant
D
Tenant
E
Serverless
Tenant
F
Provisioned compute
AWS
US_EAST_1
AWS
AP_SOUTHEAST_1
Secure
Place critical customers on dedicated Postgres compute for performance isolation and security
Low latency
Place customer's data and vector embeddings in different region for low latency
Flexibility
Move tenants between serverless and provisioned compute with no downtime or application changes
Tenant-level branching, backups, schema migration, and insights
Fine grained
branching
DB level and tenant-level branching
Branch production data for testing
Reproduce customer issues by branching specific tenant data
May
June
July
Aug
Sep
Oct
Instant customer
dashboard
1,950
1,350
650
0
Track growth of customers, embeddings and queries
Dive into per customer metrics
Manage user profiles for each customer
Automated schema migration across tenants
Execute DDL once across tenants
Schema migrations are fully atomic
Integrates with existing Postgres tooling
Tenant-level backups for instant restores
Execute DDL once across tenants
Schema migrations are fully atomic
Integrates with existing Postgres tooling
Resources
BlogCopyright © 2024 Nile