Skip to content

HeliosDB GraphRAG HTAP - Complete User Guide

HeliosDB GraphRAG HTAP - Complete User Guide

Version: 1.0


Table of Contents

  1. Introduction
  2. Core Concepts
  3. Cypher Query Language
  4. GQL Support
  5. HTAP Architecture
  6. Performance Tuning

1. Introduction

What is GraphRAG HTAP?

HeliosDB GraphRAG HTAP combines:

  • Graph Database: Native property graph with Cypher and GQL support
  • Vector Database: Integrated embeddings for semantic search
  • RAG Framework: Built-in Retrieval-Augmented Generation
  • HTAP Engine: Hybrid Transactional/Analytical Processing

Key Benefits

  • 10x Faster: Outperforms Neo4j + VectorDB combinations
  • Unified Platform: Single system vs. fragmented architecture
  • operationally hardened: WAL, backup/restore, replication, PITR
  • ACID Compliant: Full MVCC with multiple isolation levels
  • Scalable: Tested with 10M+ nodes, 100M+ edges

Use Cases

  1. Knowledge Graphs with LLM Integration

    • Build intelligent chatbots with graph-backed knowledge
    • Implement RAG pipelines with relationship-aware retrieval
    • Combine structured and semantic search
  2. Real-Time Analytics

    • OLTP queries for user interactions
    • OLAP queries for business intelligence
    • Automatic routing based on query complexity
  3. Graph Machine Learning

    • Node/edge embeddings with graph structure
    • Community detection and influence analysis
    • Recommendation systems with graph context

2. Core Concepts

2.1 Property Graph Model

HeliosDB uses the property graph model:

Nodes (vertices):

  • Unique ID
  • Label(s)
  • Properties (key-value pairs)

Edges (relationships):

  • Unique ID
  • Source and target nodes
  • Type/label
  • Weight (for weighted graphs)
  • Properties

Example:

(:Person {name: "Alice", age: 30})-[:KNOWS {since: 2020}]->(:Person {name: "Bob"})

2.2 MVCC (Multi-Version Concurrency Control)

Every modification creates a new version.

Isolation Levels:

  • Read Committed: See committed changes
  • Repeatable Read: Consistent snapshot
  • Serializable: Full serializability (with conflict detection)

2.3 HTAP Query Routing

Queries are automatically routed:

OLTP (low latency):

  • Point queries (single node/edge lookup)
  • Short paths
  • Small result sets

OLAP (high throughput):

  • Aggregations (COUNT, AVG, SUM)
  • Long traversals
  • Graph algorithms

Hybrid:

  • Mixed workloads
  • Adaptive execution

3. Cypher Query Language

3.1 Basic Queries

MATCH: Find patterns

-- Find all persons
MATCH (p:Person) RETURN p
-- Find friends
MATCH (a:Person)-[:KNOWS]->(b:Person)
RETURN a.name, b.name
-- Variable-length paths
MATCH (a:Person)-[:KNOWS*1..3]->(b:Person)
WHERE a.name = 'Alice'
RETURN b.name

CREATE: Insert data

-- Create node
CREATE (p:Person {name: 'Charlie', age: 25})
-- Create relationship
MATCH (a:Person {name: 'Alice'}), (b:Person {name: 'Bob'})
CREATE (a)-[:KNOWS {since: 2020}]->(b)

UPDATE: Modify data

-- Update properties
MATCH (p:Person {name: 'Alice'})
SET p.age = 31, p.city = 'NYC'
-- Add label
MATCH (p:Person {name: 'Alice'})
SET p:Employee

DELETE: Remove data

-- Delete relationship
MATCH (a:Person)-[r:KNOWS]->(b:Person)
WHERE a.name = 'Alice'
DELETE r
-- Delete node (and relationships)
MATCH (p:Person {name: 'Charlie'})
DETACH DELETE p

3.2 Advanced Cypher

Aggregations:

-- Count nodes
MATCH (p:Person) RETURN count(p)
-- Average age
MATCH (p:Person) RETURN avg(p.age)
-- Group by
MATCH (p:Person)
RETURN p.city, count(p), avg(p.age)

Ordering and Limiting:

MATCH (p:Person)
RETURN p.name, p.age
ORDER BY p.age DESC
LIMIT 10
SKIP 5

Conditional Logic:

MATCH (p:Person)
RETURN p.name,
CASE
WHEN p.age < 18 THEN 'Minor'
WHEN p.age >= 18 AND p.age < 65 THEN 'Adult'
ELSE 'Senior'
END AS ageGroup

Subqueries:

MATCH (p:Person)
WHERE EXISTS {
MATCH (p)-[:KNOWS]->(:Person {city: 'NYC'})
}
RETURN p.name

3.3 Performance Tips

  1. Use LIMIT Early:
-- Good
MATCH (p:Person)
WHERE p.age > 18
RETURN p
LIMIT 10
-- Less efficient
MATCH (p:Person)
RETURN p
WHERE p.age > 18
LIMIT 10

4. GQL Support

HeliosDB supports ISO GQL (Graph Query Language), the new standard.

4.1 Basic GQL Queries

-- Select nodes
SELECT *
FROM GRAPH myGraph
MATCH (p:Person)
WHERE p.age > 18
-- Path queries
SELECT p.name, f.name
FROM GRAPH myGraph
MATCH (p:Person)-[:KNOWS]->(f:Person)
WHERE p.name = 'Alice'

4.2 GQL vs Cypher

FeatureCypherGQL
StandardIndustryISO Standard
SyntaxMATCH-basedSELECT-based
Learning CurveLowMedium
CompatibilityNeo4j-likeSQL-like

When to use GQL:

  • You prefer SQL-style syntax
  • Need ISO standard compliance
  • Working with tools expecting GQL

When to use Cypher:

  • Migrating from Neo4j
  • Prefer graph-native syntax
  • Shorter, more concise queries

5. HTAP Architecture

5.1 How HTAP Works

Query
|
v
+---------------+
| Query Router |
+---------------+
/ \
/ \
v v
+----------+ +----------+
| OLTP | | OLAP |
| (Row) | | (Column) |
+----------+ +----------+
\ /
\ /
v v
+-------------+
| MVCC Storage|
+-------------+

6. Performance Tuning

6.1 Performance Targets

Indicative figures, measured on representative fixtures; reproduce on your own hardware.

MetricTargetTypical
Simple query latency<10ms2-5ms
Complex query latency<100ms30-80ms
Throughput (cached)1000 QPS2000+ QPS
Throughput (uncached)500 QPS800 QPS
Node insertion<5ms1-3ms
Transaction commit<10ms3-7ms

Conclusion

This user guide covers the essentials of HeliosDB GraphRAG HTAP. For more information:

Production Checklist:

  • Configure appropriate index strategy
  • Enable WAL and configure checkpointing
  • Set up backup schedule
  • Configure replication for HA
  • Implement monitoring and alerting
  • Performance test with production workload
  • Review security configuration

Version: 7.0.0