If you could use foreign data wrapper to connect https://github.com/tds-fdw/tds_fdw then you can skip the migration back and for to CSV.
You could even do partial migrations if needed (it could impact some queries' speed though).
Pablo
On Fri, May 3, 2019 at 6:37 AM Adrian Klaver <adrian.klaver@xxxxxxxxxxx> wrote:
On 5/3/19 4:56 AM, Matthias Apitz wrote:
>
> Hello,
>
> We're investigating the migration of our LMS (Library Managment System)
> from Sybase ASE 15.7 to PostgreSQL 10.6. The used database in field have
> around 400 columns, some of them are also containing BLOB (bytea) data.
> The DB size vary upto 20 GByte. The interfaces contain any kind of
> language one could imagine :-) ESQL/C, JDBC, Perl DBD, ...
>
> Re/ the migration of the data itself, are there any use case studies
> which could we keep in mind? We plan to export the tables with our own
> tool which produces CSV with delimiter '|' (and \| if the char | is in
> char columns too) and with hex representation of the BLOB data. This seems
> to fit nicely with PostgreSQL's COPY command.
You might want to also take a look at:
A PostgreSQL foreign data wrapper to connect to TDS databases (Sybase
and Microsoft SQL Server)
https://github.com/tds-fdw/tds_fdw
>
> Any known pitfalls?
>
> Btw: We're investigating MySQL too, but this seems to be from the list
> now for not having an ESQL/C interface.
>
> Regards
>
> matthias
>
--
Adrian Klaver
adrian.klaver@xxxxxxxxxxx
