Best for dev/test
Hourly
1.5× daily rate
- Spin up a TiDB cluster for load testing, destroy after results
- Evaluate HTAP performance before committing to monthly
- No minimum term per node type
TiDB is a distributed NewSQL database that handles OLTP and OLAP workloads simultaneously — no ETL pipeline, no separate data warehouse, no Hadoop. MySQL-compatible, Raft-based financial-grade HA, and horizontal scale to 512 nodes and PB-level storage.
Distributed architecture built for the scale problems that single-node databases can't solve — measured, verified, and deployed in financial institutions across the region.
TiDB doesn't replace your database — it replaces your database AND your data warehouse, while running MySQL-compatible SQL throughout.
Migrate from MySQL to TiDB without modifying a single line of application code. MySQL clusters using sharding middleware can be migrated in real-time using TiDB's synchronization tools — same SQL, same drivers, same ORM.
TiKV handles row-based OLTP writes while TiFlash (column store) serves real-time analytical queries — from the same data, in the same cluster. No ETL, no replication lag, no data warehouse to build and maintain.
Add TiDB compute nodes or TiKV storage nodes independently to increase throughput or capacity as needed. No complex resharding, no application downtime, no migration — expansion is fully transparent to applications.
Unlike master-slave replication where failover risks data loss, Raft majority election provides 100% strong data consistency guarantees. RPO=0 and RTO≤30s with automatic failover — no manual intervention, no risk of split-brain.
TiDB supports the complete ACID standard across all distributed nodes — not eventual consistency, not partial ACID. Multi-table, multi-node transactions work exactly as you'd expect from a single-node RDBMS.
Serverless mode auto-scales compute and storage based on actual load — no cluster configuration required. Three tiers: Trial (free), Lite (fixed small cluster), and Flagship (auto-scaling, production-ready). Focus entirely on your application.
The traditional approach forces you to maintain two separate systems — an OLTP database for transactions and an OLAP warehouse for analytics, connected by expensive ETL pipelines. TiDB eliminates this entirely.
When you write a transaction to TiDB, it goes to TiKV (row-based storage). TiFlash asynchronously replicates the data in columnar format. Your analytics queries automatically route to TiFlash — getting columnar scan performance without any ETL, without any data pipeline, and without any replication lag waiting period.
For enterprises currently running ETL + Hadoop or ETL + data warehouse: TiDB's HTAP eliminates that entire architecture. Business data synchronised from multiple source systems can be aggregated and reported on in real time — T+0, not T+1.
Learn About TiFlash →Unlike master-slave replication where failover can lose committed transactions, TiDB's Raft majority election guarantees that data is committed only when a majority of replicas acknowledge it. No data loss is possible from a single-node failure.
Choose fully customizable dedicated clusters for maximum control, or TiDB Serverless for zero-configuration deployment that auto-scales to your workload.
When your data volume exceeds what a single MySQL node can serve — and your analytics needs can't wait for nightly ETL — TiDB is the upgrade path.
| Capability | NubexCloud TiDB | MySQL (Single Node) | MySQL + Data Warehouse |
|---|---|---|---|
| MySQL SQL compatibility | ✓ 100% compatible | ✓ Native | ✓ Native |
| Horizontal scale | ✓ 512 nodes, PB storage | ✗ Single node limit | ~ Complex sharding |
| HTAP (OLTP + OLAP) | ✓ TiKV + TiFlash, one cluster | ✗ OLTP only | ~ Separate system + ETL |
| Real-time analytics | ✓ T+0, no ETL delay | ✗ Not supported | ✗ T+1 ETL delay |
| ACID distributed transactions | ✓ 100% across all nodes | ✓ Single node | ✗ Cross-system not ACID |
| HA without data loss | ✓ Raft · RPO=0 | ~ Master-slave · Risk of loss | ~ Depends on config |
| Auto-failover | ✓ <30s, no intervention | ~ Manual or semi-auto | ~ Complex multi-system |
| Infrastructure complexity | ✓ One managed system | ✓ Simple | ✗ Two systems + ETL ops |
| Storage cost at PB scale | ✓ Efficient, one cluster | ✗ Impractical | ✗ Double storage cost |
TiDB is designed for problems that traditional databases solve with complexity — sharding middleware, ETL pipelines, and separate OLAP systems. TiDB handles them natively.
TiDB Serverless billing adjusts automatically with your workload. Dedicated clusters use the same hourly, monthly, or annual model as all NubexCloud databases.
1.5× daily rate
Standard monthly rate
Pay 10 months, get 12
TiDB Server nodes, TiKV storage nodes, and TiFlash column-store nodes are each billed separately by their CPU, memory, and disk specification. This allows you to scale compute and storage independently and pay only for what you use.
MySQL-compatible. Distributed to PB scale. HTAP with TiFlash. Raft-based RPO=0. Deploy TiDB Serverless in minutes or configure a dedicated cluster to your exact specifications.