Linked Data Fragments: using client side intelligence to query the web of data

I have a work around for slow or sometimes unavailable SPARQL end points: I use a caching layer on my query clients. For the key I use the original SPARQL query and the key values are the results in either JSON form or language native form. Usually I use JSON stored in Postgres.

This makes unit tests possible and generally speeds up interactive development of systems where linked data sources are used.