What format should I use for scanned invoices in SQL Server?

Question

Hello!

My client currently scans invoices and similar documents into PDF format, and stores them in the file system. I suggested to him that we store the scans in their custom management system's SQL Server to remove one of the 'moving parts' of their system.

My original idea was to store them in a lossless graphics format, but the accountant is worried that this will not be good enough for the IRS or the courts, should the documents be needed for any purpose. Do you know if providing the ability to export these scanned docs as PDFs on demand would be acceptable for these purposes, or should I store the docs in PDF in the database to start with?

Thanks for your advice!

Answer 1

Any legal advice that you get on SO (or anywhere on the internet) is worth every cent you pay for it. Bear that in mind.

If you're planning on storing a graphical representation of the document, why not simply store the PDF instead? All of the philosophical arguments regarding storing large binary objects in the database aside, I don't see a real advantage to storing an image vs. a PDF, especially if you know the PDF is acceptable.

Answer 2

Purely from a technical point of view (I don't think you'll get any legal advice here!)....

If using SQL 2005 or lower, I would only store the path to the PDF in the database, not the file system. SQL Server != File System.

If using SQL 2008, you could store it as a FILESTREAM - which (quote):

FILESTREAM integrates the SQL Server Database Engine with an NTFS file system by storing varbinary(max) binary large object (BLOB) data as files on the file system.

Answer 3

<IANAL but="I've had to give testimony">: If the accountant is under the impression that plain old PDFs are safe from alteration, that's unfortunately a common misconception.

Once you get into legal issues, the documents you present during discovery/litigation will simply be accepted on face value. I know this sounds wrong, but most parts of the legal system rely on other checks and balances (like additional charges for perjury and fraud) to discourage falsification of documents.

Only if there is a question of fraud or chain of custody or anything like that, then those kind of forensic data investigations occur and it would be nice if you have the documents watermarked or digitally signed with a strong signature mathod and an audit trail or similar to prove that the data has not been altered. This gets into much deeper aspects like administrator sights on the server versus data change roles and key management. It also becomes a question of cost-benefit, like all security questions.</IANAL>