Room: Baltic I
At parcIT we are developing algorithms for the credit industry. To develop and validate their algorithms, our analysts require a large pool of anonymized real world data.
This data pool had been spread on 3 smaller MS-SQL Databases before and is now consolidated to one Postgres DB that has currently grown to 25TB on high performance storage and will be enlarged up to 4 times that size.
The original setup of this Postgres Cluster hit surprisingly many pitfalls and let to a massively instable system with some security concerns. Together they had the potential to lead to a failure of the whole project. It took us several weeks to track the problems down even though all of them were more or less easy to avoid.