Classify

enaio® uses the smart FIX component from Insiders to classify documents. Aside from the restrictions and guidelines set out by the software maker, the following recommendations are proven best practices:

smart FIX requires additional databases located on an MSSQL server. System sizing will vary depending on the daily volume of incoming data. At least 1 GB RAM and one CPU on the MS SQL server should be reserved for smart FIX. The classification component is deployed on a dedicated host and requires resources in line with the volume of documents. As a rough guide, it can be assumed that 20 to 30 seconds will be required to extract one A4 page (150 dpi) if you are using a 3 GHz CPU. This is a worst-case scenario, which is a reasonable assumption considering the scanned originals to be processed, the quality of the scanned originals, and other use of infrastructure. Processing typically takes less time. The specified RAM usage is 1 GB per CPU core.

The example above assumes a highly constant flow of incoming documents. If large batches of documents are received by the system during certain weeks or months, the requirements will change based on the demands in terms of processing speed. As a further example, it is assumed that the documents generated during a month (20,000) need to be processed in a single day (eight hours).

In short: 20,000 pages/eight hours/60 minutes = 41.7 pages/minute. 20 CPU cores will in turn be needed for processing by the classification system.

The key variable is thus the desired processing speed, taking into account the intermittency of incoming documents. Scaling must be carried out on this basis.