HeliosProxy v1.0.0 Release Notes

HeliosProxy v1.0.0 Release Notes

Release Date: January 2026 Branch: heliosproxy

---

Overview

HeliosProxy v1.0.0 marks the transformation from a basic connection router into an intelligent, AI-ready database proxy for modern applications. This release includes 14 major features spanning connection pooling, resilience, enterprise capabilities, and distributed caching.

---

Highlights

• Connection Pooling Modes - Session, Transaction, and Statement pooling for 100x connection efficiency
• Intelligent Caching - Query caching with TTL, invalidation, and distributed L1/L2/L3 cache tiers
• AI/Agent Native - First-class support for RAG pipelines, agentic workflows, and vector operations
• Enterprise Ready - Multi-tenancy, authentication proxy, rate limiting, and WASM plugin extensibility
• High Availability - Circuit breaker, replica lag-aware routing, and automatic failover

---

New Features

Phase 1: Foundation

1. Connection Pooling Modes (Feature Flag: pool-modes)

Three pooling modes to reduce backend connection overhead:

Mode	Behavior	Best For
Session	1:1 client-to-backend mapping	Legacy apps, prepared statements
Transaction	Return after COMMIT/ROLLBACK	Web apps, microservices
Statement	Return after each statement	Read-heavy, connection-starved envs

[pool_mode]
mode = "transaction"
max_pool_size = 100
min_idle = 10
idle_timeout_secs = 600

2. Query Caching

Transparent query result caching with configurable TTL and invalidation strategies:

• LRU eviction with memory limits
• Table-based invalidation on writes
• Cache bypass via SQL hints

[cache]
enabled = true
max_memory_mb = 512
default_ttl_secs = 60

3. Query Routing Hints

Control query routing via SQL comments:

SELECT /* helios:route=primary */ * FROM orders;
SELECT /* helios:route=standby */ * FROM analytics;
SELECT /* helios:cache=skip */ * FROM live_data;

4. Replica Lag-Aware Routing

Automatically route queries to replicas within acceptable lag thresholds:

[routing]
max_replica_lag_ms = 100
lag_check_interval_secs = 1

Phase 2: Resilience & Control

5. Rate Limiting & Query Throttling

Protect backends from overload:

[rate_limit]
queries_per_second = 1000
burst_size = 100
per_user = true

6. Circuit Breaker Pattern

Prevent cascade failures with automatic circuit breaking:

• Closed: Normal operation
• Open: Reject requests when errors exceed threshold
• Half-Open: Probe for recovery

[circuit_breaker]
failure_threshold = 5
recovery_timeout_secs = 30

7. Query Analytics & Slow Query Log

Built-in query performance monitoring:

• Query frequency and latency histograms
• Slow query logging with configurable threshold
• No pg_stat_statements required

[analytics]
enabled = true
slow_query_threshold_ms = 100

Phase 3: Enterprise Features

8. Multi-Tenancy Support

Isolate tenants with dedicated pools and quotas:

[[tenants]]
id = "tenant_a"
max_connections = 50
rate_limit = 500
databases = ["tenant_a_db"]

9. Authentication Proxy

Proxy-level authentication with multiple backends:

• LDAP integration
• JWT validation
• Certificate-based auth

10. Query Rewriting

Transform queries transparently:

• Schema prefixing for multi-tenancy
• Query normalization
• Deprecated syntax migration

Phase 4: Differentiation & Innovation

11. WASM Plugin System (Feature Flag: wasm-plugins)

Extend proxy behavior with WebAssembly plugins:

[[plugins]]
name = "audit_logger"
path = "/plugins/audit.wasm"
config = { log_level = "info" }

Plugin hooks:

• on_query_start
• on_query_complete
• on_connection
• on_error

12. GraphQL-to-SQL Gateway (Feature Flag: graphql)

Automatic GraphQL endpoint generation:

[graphql]
enabled = true
endpoint = "/graphql"
introspection = true

13. Schema-Aware Routing

Route queries based on table/schema metadata:

[[schema_routes]]
tables = ["events", "logs"]
target = "analytics_replica"

[[schema_routes]]
tables = ["orders", "payments"]
target = "primary"

Phase 5: Distributed Intelligence

14. Helios-DistribCache (Feature Flag: distribcache)

Multi-tier intelligent caching for AI workloads:

Cache Tiers:

• L1: In-process (microseconds)
• L2: Shared memory (sub-millisecond)
• L3: Distributed cluster (milliseconds)

Workload Classification:

• OLTP (transactional)
• OLAP (analytical)
• Vector (similarity search)
• AI/RAG (embedding pipelines)

AI-Specific Optimizations:

• Embedding prefetch for RAG pipelines
• Conversation context caching for agents
• Branch-aware caching for experiments

[distribcache]
enabled = true
l1_size_mb = 64
l2_size_mb = 512
l3_nodes = ["cache1:6379", "cache2:6379"]

[distribcache.ai]
embedding_prefetch = true
conversation_ttl_secs = 3600

---

Configuration

Minimal Configuration

listen_address = "127.0.0.1:5433"
admin_address = "127.0.0.1:9090"

[pool_mode]
mode = "transaction"
max_pool_size = 100

[[nodes]]
host = "127.0.0.1"
port = 5432
role = "primary"

Full Configuration Reference

See docs/guides/configuration-reference.md for complete configuration options.

---

Build Options

# Minimal build
cargo build --release -p heliosdb-proxy

# With connection pooling modes
cargo build --release -p heliosdb-proxy --features pool-modes

# With distributed caching
cargo build --release -p heliosdb-proxy --features distribcache

# Full feature build
cargo build --release -p heliosdb-proxy --features pool-modes,distribcache,wasm-plugins,graphql

---

Migration Guide

From Direct HeliosDB Connections

1. Deploy HeliosProxy between application and HeliosDB
2. Update connection strings to point to proxy address
3. Start with mode = "session" for compatibility
4. Gradually migrate to mode = "transaction" for efficiency

From PgBouncer

HeliosProxy supports similar pooling modes with enhanced features:

PgBouncer	HeliosProxy	Notes
session	session	Direct mapping
transaction	transaction	Direct mapping
statement	statement	Direct mapping
N/A	Query caching	New capability
N/A	AI workload routing	New capability

---

Breaking Changes

None - this is the initial release.

---

Known Limitations

1. Statement pooling does not support server-side prepared statements
2. GraphQL gateway requires schema introspection enabled on backend
3. WASM plugins have 100ms execution timeout per hook
4. DistribCache L3 requires external Redis-compatible cluster

---

Performance Benchmarks

Scenario	Direct Connection	HeliosProxy (Transaction Mode)
Connection overhead	50-200ms	<1ms (pooled)
Simple SELECT	0.5ms	0.6ms
Cached SELECT	N/A	0.05ms
Connections @ 1000 QPS	1000	10-50

---

Documentation

• Roadmap
• Feature Specifications
• Tutorials
  • pool-modes-tutorial.sh
  • ha-pool-failover-tutorial.sh
  • distribcache-tutorial.sh

---

Contributors

HeliosDB Team

---

What’s Next

Future considerations for HeliosProxy:

• Adaptive pool sizing based on load
• Query plan caching
• Cross-datacenter cache replication
• ML-based query routing optimization
• Protocol support for MySQL wire protocol