Ready to Start?Discover our quick start guide and build your first application in 5-10 minutes
Quick Start Guide
Download IgniteARCHITECTURAL BACKBONE
CORE FEATURES
DATA PLACEMENT & PROCESSING
OPERATIONAL EXCELLENCE
ADDITIONAL USE CASES
Download IgniteARCHITECTURAL BACKBONE
CORE FEATURES
DATA PLACEMENT & PROCESSING
OPERATIONAL EXCELLENCE
ADDITIONAL USE CASES
Apache Ignite
The Architectural Backbone
Two foundational choices that make memory-first distributed computing possible
Two foundational architectural choices enable everything Apache Ignite delivers. These aren't implementation details. They're the backbone that makes memory-first distributed computing possible without sacrificing consistency or availability.
MVCC: Multi-Version Concurrency Control
The Fundamental Problem
How do you serve reads and writes concurrently without blocking? Traditional locking approaches force a choice: either readers block writers, writers block readers, or you accept eventual consistency.
The Solution
MVCC solves this by creating a new version for every write, tagged with a timestamp. Readers work against stable snapshots while writers create new versions. Readers and writers never block each other.
How It Works
Every row exists as a version chain. Each write adds a new version with begin and commit timestamps. Transactions read from a snapshot determined by their start time. The system maintains multiple versions until older snapshots complete.
What This Enables
Snapshot isolation (REPEATABLE_READ) without traditional locking overhead. Analytical queries execute on stable snapshots while transactional updates proceed at full speed. Long-running reads don't block operational writes.
Distributed Replication
Consensus-Based Replication
Making distributed state changes safe requires consensus. Apache Ignite uses Raft consensus for replication. Each partition forms a Raft group with a leader and followers.
How Leader Election Works
When a leader fails, the remaining nodes elect a new leader through the Raft protocol. Leader election completes in milliseconds. The new leader continues serving requests without data loss.
WAL-Free Architecture
Apache Ignite uses Raft consensus for durability instead of traditional write-ahead logs. The leader processes writes and replicates through the distributed consensus log. No local WAL needed.
Durability and Availability
Data is durable once replicated to a majority of replicas. The cluster remains available as long as a majority of replicas are reachable. This provides both durability and availability without traditional logging overhead.
What This Foundation Enables
ACID Transactions
MVCC provides snapshot isolation without blocking readers. Distributed replication ensures durability across partitions. Full ACID transactions work across any number of partitions.
Memory-First Storage
Raft-based replication provides durability without traditional write-ahead logging. AIPERSIST uses this for sub-millisecond persistence. AIMEM leverages version chains for microsecond-level access.
Concurrent Workloads
MVCC allows analytical queries and operational writes to run simultaneously. Long-running aggregations work on stable snapshots while real-time updates proceed at full throughput.
High Availability
Raft leader election ensures continuous operation during failures. Automatic failover completes in milliseconds. The cluster remains available as long as a majority of replicas are reachable.
Ready to Start?
Explore FeaturesSee how these architectural foundations enable all Apache Ignite capabilities
Features Overview