The MySQL box is being retired, PostgreSQL is already provisioned, and the database is 40 GB. mysqldump gives you a 40 GB text file that no server will happily ingest, the type mapping (tinyint(1) → boolean, datetime → timestamptz, every enum you ever wrote) is still hand work, and the cutover has to happen while both databases are live, in both directions if the rollback is needed next week.
That is the job this category exists for, and it is why the searches are so specific: mysql to postgresql, db2 to sql server migration, oracle to mysql migration, dbf to sql converter. Nobody wakes up wanting “a database tool”; they want one pair, done, without a maintenance window spent writing scripts.
These are the tools that come up for those queries, what each one is genuinely good at, and — since the honest answer depends on which pair you need — how to pick without paying for the direction you will never use.
Disclosure: the database tools at the end of this article are our own products. They get their own section after the alternatives, and they are in the table so you can compare everything at once.
How to judge a database conversion tool
The criteria that decide whether a tool survives a real migration, not the demo database:
- Does it cover your pair, in the direction you need? Every vendor converts to their favourite engine; far fewer convert both ways, and fewer still cover the pairs nobody writes blog posts about (DB2 → SQL Server, dBASE → SQLite). Check the engine list, not the marketing count.
- What does it need on the machine? “Native drivers” is the difference between a tool that just runs and an afternoon installing an ODBC driver, a DSN per database and the right bitness of each.
- Can you control the mapping? Column types, nullable/default, keys and indexes, and what happens to a table that already exists — append, upsert, truncate, skip. Automatic mapping is a start; automatic mapping you cannot override is a bug report waiting for the first
enum. - What happens at a million rows? A desktop tool that stages everything in memory works on the sample database. Streaming/batched inserts and a per-table row counter are what make a 40 GB migration finish instead of dying at 60%.
- Can you re-run it? A cutover is almost never one run: dry run, full run, incremental catch-up, then the rollback you hope not to use. Saved sessions, a CLI and scheduling turn that into a repeatable job instead of a wizard you re-click.
- Does the data stay local? Database contents are the most sensitive data most companies own. Cloud “migration services” are legitimate, but they need credentials to both databases and a network path to their servers.
- What is the price model? Per-database-pair perpetual licences, subscription, or tiers — and does the tier you can afford include the pair you actually need?
The comparison
| Tool | Best for | Price | Watch out |
|---|---|---|---|
| Full Convert | the widest engine list in the category, including ODBC/legacy sources | subscription: Pro $499/quarter or $999/year, Ultimate $4,999/year | priced as a subscription; a perpetual option is listed but its price is rendered dynamically (unverified 2026-11-06) |
| DBConvert | one well-known pair at a time (MySQL↔PostgreSQL, SQL Server→MySQL, …) | from US$199 per database pair; DBConvert Studio US$599; licence never expires, 1 year of updates | per-pair licences add up across a migration programme |
| DBeaver (Community) | open-source DB client you probably already have | free (Apache-2.0); paid PRO/Enterprise editions exist | the data transfer wizard is not a migration tool: no saved job, no scheduler, limited cross-engine type control |
| Withdata DBCopier | any pair among 8 engines — SQL Server, MySQL/MariaDB, PostgreSQL, SQLite, Oracle, Access, DB2, DBF/dBase — with sessions, upsert and a CLI | $129 one-time, 30-day full trial | desktop install; 8 engines, not 40 |
The free scriptable baseline (pgloader, ora2pg, mongoexport) is in its own section below.
Full Convert — Spectral Core
The reference tool for this category, and the one that shows up in every “mysql to postgres” thread. Its own homepage claims over 40 database engines supported directly with no ODBC, which is the widest coverage you can buy for desktop migration, and it has matured for the awkward pairs (Interbase, Firebird, Paradox, DBF, and friends) that nobody else keeps alive.
The price model has moved to subscription: the order page lists Full Convert Pro Quarterly at $499/quarter, Pro Annual at $999/year, and Ultimate Annual at $4,999/year. A “Perpetual” tab exists on that page, but the numbers behind it are rendered by script, so on 2026-11-06 we could only verify the three subscription prices. For a one-off migration that is a real decision: $999/year against a $129 one-time licence is the whole argument in one line.
DBConvert — per-pair licences
DBConvert covers the well-travelled pairs as separate products (MySQL↔PostgreSQL, SQL Server→MySQL, Oracle→PostgreSQL…), with DBConvert Studio as the multi-route version. Their pricing page states it plainly: “DBConvert starts at US$199 per database pair, DBConvert Studio at US$599. Every license includes a year of free updates.” The licences never expire — so the question is not subscription fatigue, it is arithmetic: two directions are two pairs, a third engine turns the matrix into five or six licences, and at that point Studio is the cheaper shape.
Where it is strong: the individual pair tools are focused and documented, and Studio’s schema/layout mapping is genuinely deep. Where to be careful: the per-pair model means “just add one more source” is a purchase, and like every vendor the price list is worth re-checking before you budget.
DBeaver Community — free, and already installed
If you already use DBeaver, its data transfer feature will move tables between two connections: right-click a table, pick the target, run. It is free (Apache-2.0), cross-platform, and good enough for “copy these four tables into staging once”.
What it is not: a migration tool. There is no saved-and-scheduled job, no per-run upsert policy to lean on for the catch-up run, no cross-engine type mapping you can review as a mapping, and no row-level progress you can quote to whoever is waiting for the cutover. For a one-off small copy it wins on price; for a 40 GB cutover it is the thing you start with and then replace.
The free baseline: pgloader, ora2pg, mongoexport
If the pipeline is already code, the free answers are real:
pgloaderloads MySQL, SQLite, CSV and DBF into PostgreSQL, with type casting rules and a migration file you can put in version control — the best free MySQL→PostgreSQL path, and PostgreSQL is the only target.ora2pgexports Oracle schemas and data to PostgreSQL or MySQL, with a configuration file that maps types, and is the standard free Oracle exit route.mongoexport/mongoimportmove collections to JSON/CSV and back — one direction of one format at a time, and no relational side at all.
All free, all scriptable, all requiring you to write, test and maintain the mapping file — and all of them assume you are comfortable being the only support line when the mapping is wrong at 2 a.m.
The Withdata database tools
Our own line-up is deliberately split by direction instead of by engine pair, which is the main structural difference from the tools above:
| Product | Direction | Price | Notes |
|---|---|---|---|
| DBCopier | database → database, any pair | $129 | v3.1.1 · 8 engines — SQL Server, MySQL/MariaDB, PostgreSQL, SQLite, Oracle, Access, DB2, DBF/dBase — table-to-table, SQL query → table, saved sessions, upsert, CLI + scheduler; Windows 32/64, Linux (.deb/.rpm), macOS 64-bit |
| DBToFile | database → CSV / Excel / JSON / TSV / TXT / XML | $75 | batch export, CLI, scheduled runs · v3.1 |
| FileToDB | CSV / Excel / JSON / XML / TSV / TXT → database | $75 | v3.7.1 · the same 8 engines as targets; CLI; free trial limited to 1,000 rows per file |
| DBLoader | bulk load, both ways | $129 | 10+ databases and 10+ file formats, built for the big file case rather than the single-table case · v3.7; trial has no time limit but limits batching and CLI |
| MongoToDB / DBToMongo / MongoToFile / FileToMongo | MongoDB ↔ relational / files | $75 ($59.95 for the file two) | MongoDB side of the same job · MongoToDB/DBToMongo v3.1.1, FileToMongo v5.2.1, MongoToFile v4.9 |
What that buys you, on this article’s queries: MySQL → PostgreSQL is one licence (not a per-pair purchase), DB2 → SQL Server is the same licence, and dBASE/DBF → SQL is covered without a second product. What it costs you: Full Convert’s 40+ engines include legacy sources we do not cover, so if your source is Paradox, Informix or Interbase, buy theirs. The honest split is coverage versus price model.
Two limits worth stating before the how-to:
- 8 engines, not “20+”. Our product pages advertise “20+ databases”; that count includes engine variants (MySQL 5.x/8.x, Oracle versions, and so on). The engine families are the eight in the table above — check yours against that list, not against the number.
- The trial is limited, the licence is not. DBCopier’s trial runs 30 days with full features and no credit card; FileToDB’s trial caps imports at 1,000 rows per file; DBLoader’s has no time limit but limits batch sizes and advanced CLI use. Test with your own table before the maintenance window.
Which one to pick
- One pair, and you already know the tool you want: DBConvert’s per-pair product if the pair is on their list ($199+, perpetual). It is the most focused option for exactly MySQL↔PostgreSQL and friends.
- Legacy source (Paradox, Interbase, Informix) or a source we do not cover: Full Convert — 40+ engines is the reason it exists, and the subscription is the price of that coverage.
- Small, one-off copy between two DBs you already admin: free — DBeaver‘s data transfer, or
pgloader/ora2pgif the target is PostgreSQL. - Your pair is DB2 → SQL Server, dBASE → SQLite, Access → Oracle, or MongoDB ↔ anything relational: a tool that does any pair among its engines for one price — DBCopier ($129) — is cheaper than a per-pair licence, and it is the only row in the table where DBF/dBase is a first-class engine rather than an ODBC afterthought.
- Files into a database, or a database into files, in bulk and on a schedule:
DBLoader/DBToFile/FileToDBrather than a migration tool; the migration tools are built for table-to-table moves.
How to migrate MySQL to PostgreSQL in four steps
The flow is the same one that needs to be repeated for the rollback, so do it as a saved session from the start.
- Connect both sides in
/db-copier/: source MySQL (host, port, user, password, or a saved connection), target PostgreSQL. No DSN, no ODBC driver to install — the tools speak the database protocols directly, and both panels can be saved into a session file. - Choose the scope: whole database, a table list, or a SQL query → table copy when you only want the last 90 days of
orders. The query path is what makes a dry run cheap. - Review the mapping — column types, keys, what happens to existing rows. For the cutover run: create tables; for the catch-up run: upsert, so yesterday’s rows are not duplicated when the job runs again.
- Run it and keep the receipts: per-table row counts on both sides, and a saved session you can re-run from the command line — that session plus a scheduler entry is the “nightly sync” and the “rollback in 20 minutes” story, without a second tool.
Frequently asked questions (FAQ)
Is there a free database conversion tool?
Yes, with a shape: pgloader (into PostgreSQL), ora2pg (Oracle out to PostgreSQL/MySQL) and mongoexport/mongoimport are free and scriptable; DBeaver Community moves tables between connections for free. What you pay for with a commercial tool is the GUI, the type mapping you can review, saved/scheduled jobs and upsert for re-runs.
Can I convert MySQL to PostgreSQL without installing ODBC drivers?
With the tools that speak native protocols (DBCopier, Full Convert’s direct engines, DBConvert), yes. ODBC is the fallback for engines a tool does not handle natively — which is the usual reason a “supported database” turns out to be supported only through a driver hunt.
How big a database can this handle?
The vendors all quote “millions of rows per minute”, and it is the right order of magnitude for local, batched copies. The practical limit is disk, indexes and network between the two servers, not RAM: a tool that streams and batches keeps memory flat, which is what you want on a 40 GB source. Test on your own table and measure before the maintenance window.
What happens to my enums, booleans and time zones?
tinyint(1) → boolean, datetime → timestamp/timestamptz, and MySQL enum → a lookup table or a check constraint: none of these are mechanical, and every tool’s automatic mapping differs. Review the mapping in a dry run, especially for datetime columns, where a missing time zone assumption silently shifts data by hours.
Do I have to upload the data to a cloud service?
No, and for customer data that should be the deciding factor. All the desktop tools here connect from your machine to your two databases; nothing is uploaded. Cloud migration services (AWS DMS, Azure DMS) are legitimate, but they need credentials for both databases and a route to the provider.
Can it copy DBF/dBase or Access into a modern database?
Yes in our case — DBF/dBase and Access (including the 32-bit mdbtools-era files) are engines in DBCopier, not ODBC afterthoughts, and that is the common “dbf to sql converter” job: dBASE tables into SQL Server, MySQL, PostgreSQL or SQLite.
Final thought
Pick by the pair, then by the price model. One known pair, permanent licence, cheapest: DBConvert per-pair. A legacy source, or a pair nobody lists: the 40+ engine tool, at subscription prices. A copy you do once between two servers you admin: pgloader or DBeaver, free. Any pair among the eight mainstream engines — and especially DB2, Access, dBASE/DBF and MongoDB on either side — one $129 licence with a session you can re-run from cron solves the same problem as the per-pair and subscription rows above, and that is the honest reason we build it that way.
Facts checked November 2026. Prices, versions and supported platforms change — verify on each vendor’s page before buying.
Sources: Full Convert order page (Pro Quarterly $499, Pro Annual $999, Ultimate Annual $4,999; perpetual prices script-rendered — checked 2026-11-06) · Full Convert (“over 40 database engines supported directly with no ODBC”, checked 2026-11-06) · DBConvert pricing (“DBConvert starts at US$199 per database pair, DBConvert Studio at US$599 … year of free updates”, checked 2026-11-06) · DBeaver Community (Apache-2.0) · pgloader / ora2pg / MongoDB Database Tools documentation · our own pages: DBCopier $129, DBToFile $75, FileToDB $75, DBLoader $129, MongoToDB $75, DBToMongo $75, MongoToFile $59.95, FileToMongo $59.95 (checked 2026-11-06).
DBCopier without the manual work
For batch jobs, large files and scheduling use the DBCopier converter — an offline desktop app for Windows, Linux and macOS. Free trial.