Since we started working in document management systems, one of the most significant challenges we face are migrations.
Our team has already migrated more than one hundred million documents for our various clients. Even still, each migration is a big challenge; we never know what we are going to find on the other side.
We have had migrations that range from file systems to custom-made document managements systems to totally obsolete document management systems such as old versions of Alfresco, Sharepoint, Zope, Nuxeo, and so many more.
Curiously, there are always problems that come up again and again that we can draw some recommendations from.
Companies always think that the quality of the data that their document managements systems have is better than what it really is. In this situation, a migration may turn into an ELT process. This is something we want to avoid at all costs. It’s recommendable that you correct the original source or, if this is not possible, make the correction in a post-migration process. Carrying out this process during the migration makes the process much more complicated and also makes quality tasks more difficult.
It’s true that there is an upshot to doing it during the migration: the process is shorter.
It is very common that the coding of the documents and metadata to be migrated is obsolete. All modern document managements systems normally work with utf8. However, it is common for the source code to be latin1 or some even older code that was used in Windows systems.
In this case, our recommendation is to take the coding into account in order to make the opportune conversions. If this isn’t done, the user experience is terrible, as you have to deal with text with weird characters.
Tackle the binaries first
Many people tackle the migration to a document management systems as if it were a database migration. They forget the most important thing: the documents.
Documents are the heaviest part when it comes to migrating to a document management systems and are the most critical part when it comes to time frames. All the migrations that we have done had a very limited time frame. Normally, a migration is linked to a production launch process and the migration is an indispensable part in order to be able to start production without missing any documents.
For this reason, it is important to tackle migration of the binaries from the beginning. Determine how long they take to transfer and draw conclusions with regards to the total time it may take to move the terabytes of existing documentation.