1.1. --The best way varies on the quantity of data and desired access. Views module can provide reasonable filtering for date ranges or Solr can have additions to allow it to return results based on a date range.
1.2 --PHP has DB2 connectors that can be installed. It is probably easiest to install the DB2 connectors for PHP, define a secondary DB in the settings file for DB2, and send requests to DB2. It would be straightforward to provide XML requests to the DB2 server as well, however, that would require additional work on the part of the DB2 site maintainers.
1.3--ame as 1.2. If by process on DB2 Server vs process on the local machine you are suggesting aggregation of results from the database, then an XML communication would be suggested for 1.2 and the current answer to 1.2 would be suggested for 1.3.
1.4 --The data would need to be surfaced – at which point it could be crawled with Nutch – or pushed into the index – requiring development on the DB2 side. Once the content is in the index, any content in the index can be retrieved via Drupal. Note that surfacing results would be atypical, however, in that there would be no “page” to navigate to from the search without pulling the content into Drupal or providing an external-facing page that can be accessed. DB2 data could also be entirely ingested into Drupal and then pushed into the Solr index.
1.5 --Issues similar to 1.4 would exist, however, pushing documents into the index would only require the installation of the Tika extension to Solr. Tika allows documents to be scanned, indexed, and retrieved via Solr. Like 1.4, once the items are in the index, they would show in results, but retrieving the file would require the file to be in Drupal, externally accessible, or retrieved via custom code in Drupal. Note that I do not believe that Tika will be able to turn images into text that can be indexed. If the goal is to have the content be indexed and searchable via keywords, it would be suggested that the documents are tied to a “content” item, fleshed out with keywords or other metadata, and served via the content item rather than directly – in this case Tika would not be necessary.
1.6 --If we are accessing the DB2 system directly, there need be little to no impact on the client’s staff. If the access is done via an api of some sort (xml), then the xml api will need to allow for this sort of flexibility up front.
1.7 --Items can be pushed into the Solr index at any time via a variety of means. Drupal does not provide a directory-scan style import of items nor does Solr independently. Nutch would allow you to scan over the available items for import into the index, however, without context the TIFF images would not be searchable in a meaningful fashion.
1.8 --We are not currently aware of any tool that will convert TIFF images into searchable PDFs, though commercial solutions may exist of which we are unaware.