Migration speed is one of the main obstacles teams hit when moving data from Oracle to PostgreSQL. CYBERTEC Migrator is built to remove that obstacle, combining parallelization and change data capture (CDC) to keep throughput high without sacrificing reliability. The product is fully Kubernetes and OpenShift ready.
In testing, the migrator has reached up to 1.5 GB / second, enough to move terabytes of data in the shortest feasible time. In practice the remaining limits are the source database, the network and the destination database — hardware quirks included. The code scales well, and Big Data companies have used it to complete their transition to PostgreSQL.
What makes the transfer fast
Professional customers demand professional performance
The performance comes from two decades of engineering experience producing optimized code. The mechanisms behind a high-speed migration are:
- Massively parallel data transfer
- Parallel index creation
- Concurrent key and constraint deployment
- Efficient data transfer based on hand-optimized, battle-tested, Big Data-capable C code
Handling faulty data without slowing down
A migration is also a question of trust, verification and consistency. The migrator therefore includes techniques for detecting faulty data and dealing with it in a controlled way — in regulated sectors such as banking and finance, correct handling of faulty data can be a legal requirement.
| Mode of operation | dry run | migrate & skip error | zero tolerance migration |
| data transfer: | simulate migration | migrate correct data | migrate data |
| handle errors: | catch faulty data | catch faulty data. migrate correct data | stop on error |
A dry run simulates the migration and collects the faulty rows that would otherwise cause problems during the real run. The result is an easy-to-use data set listing every error and violation detected by PostgreSQL.
When a migration is fault-tolerant — typically because only a handful of unimportant rows are broken — the tool performs a hyper-speed migration of all data and returns the list of faulty entries for later inspection.
Zero-tolerance migrations stop altogether once a faulty row is found: data is migrated, but the process halts immediately at the first broken piece of data. These modes let teams choose how strictly to enforce data quality without giving up migration speed.
Compared with ora2pg
The advantages over ora2pg are not limited to raw throughput. Alongside high-speed transfer, intelligent CDC reduces downtime, giving an almost zero-downtime migration that is unique in this space. The result is:
- High data transfer during migration
- Lowest possible switchover times
A live demonstration of high speed data migration can be requested through the contact page.
