(This is a re-cap of a series of slides I presented last year (2012) at the XBRL Data Forum – Data Structures, Austin, Texas)

126 MB – The size of the instance document in this filing: http://www.sec.gov/Archives/edgar/data/65984/000006598413000050/0000065984-13-000050-index.htm]

The size of the instance document is much larger now because more and more companies are tagging the granular data in the disclosures. The size of the instance documents from the REITs and banks could exceed 10 MB.

The large size of the instant documents poses a challenge in consumption:

  1. Time to download
  2. Time to process
  3. Memory
  4. Network bandwidth
  5. Storage

It would be nice if the instance document can be broken up into 3 documents:

  1. One for the TextBlock data (such as the HTML contents for the disclosures);
  2. One for the Non-Textblock data (mostly numeric data);
  3. One for the Context data that can be referenced by the two above (such as legal entity, start_date,end_date, units, measures, etc.).

Majority of the users are mainly interested in the numeric data. They could download numeric data which could be 500 KB instead of 5 MB. More important, Web applications can then process smaller documents on the fly.

[the end]