Exactly, I have done most of my work on legacy (think 10-15 years worth of data, millions of documents that were spawned in countless application versions, without any schema tracking whatsoever - everything is implicit in the document itself) Lotus Domino applications.
Since there is no explicit database schema in these types of databases, what you didn't do at write, you have to do at read. And usually you want to use the latest view or representation of data, what do you do with data that wasn't there ten versions ago. What do you do about data fields of wrong format? What do you do with data that is "orphaned" and cannot be referenced to other data. Yet it still is data and still is important.
Don't get me wrong, I Love NoSQL and I like to use it. I just have enough experience with it to know that it is definitely not a silver bullet.
By the way: If someone is looking to hire a guy who is not afraid of tackling this kind of issue, contact me. I have plenty of experience with coercion of non relational data into a form suitable for analysis.
Since there is no explicit database schema in these types of databases, what you didn't do at write, you have to do at read. And usually you want to use the latest view or representation of data, what do you do with data that wasn't there ten versions ago. What do you do about data fields of wrong format? What do you do with data that is "orphaned" and cannot be referenced to other data. Yet it still is data and still is important.
Don't get me wrong, I Love NoSQL and I like to use it. I just have enough experience with it to know that it is definitely not a silver bullet.
By the way: If someone is looking to hire a guy who is not afraid of tackling this kind of issue, contact me. I have plenty of experience with coercion of non relational data into a form suitable for analysis.