Performance problems with FHIR XML schema

The low performance of the FHIR pipeline gave us some headaches. It was significantly slower than the other pipeline although being very similar in the its structure. Turned out that the result validations with the official FHIR schema was the culprit. Initially we used “fhir-all” in <p:validate-with-xml-schema>, but that was extremely slow. Turns out that schema document just imported about 140 other schema documents, and each of them importing the same schema. Luckily there is an other official schema document “fhir-single” which contains a complete schema document. Using this schema document did the trick and we are now happy with the pipeline's performance.