0

In the company I work for data administrators dislike using large data types. It doesn't matter the DBMS we are working with. As an alternative the ask developers to store data in file system or some kind of storage. If those types exist so why can't developers use it to store metadata? Are there any plausible reason?

EDIT: In my case I need to store system configuration. I mean very small data. 95% of records are about 50 kbytes and sometimes about 128 kbytes. No images just information. There is a rule of thumb here that you shall not use those data types at all and I cannot understand. Maybe folks here can explain why.

I believe that this happens on other companies also so please don't close my question.

Eduardo
  • 5,645
  • 4
  • 49
  • 57
  • 1
    It makes their lives easier: less GB to take care of, less work for them. *I even had to deal with a DBA that wouldn't allow CLOBS because he used SQL*Plus to inspect tables and data didn't show un in plain `SELECT * FROM foo` queries.* – Álvaro González Aug 31 '15 at 10:37
  • 2
    You might want to show the SQL Server DBAs this Microsoft research paper: http://research.microsoft.com/apps/pubs/default.aspx?id=64525 Most probably the results wouldn't be _that_ different for other DBMS –  Aug 31 '15 at 10:39
  • 2
    It's also a matter of cost: DB storage is usually the most expensive due to high performance and high availability requirements which means "misusing" a relations dbms for a document storage from their point of view. – Filburt Aug 31 '15 at 10:41
  • Also: http://stackoverflow.com/q/8407372/330315 and http://stackoverflow.com/q/3748/330315 –  Aug 31 '15 at 11:13

1 Answers1

2

Are there any plausible reason?

There is no definitive answer to that. For each situation there is a best place to store the data, and sometimes that can be the database, but definitely not all of the time.

There are several reasons not to store binary data in a database, but there are also several reason not to store it anywhere else. These vary for each situation, so you would need to consider each sitation separately to get the best decision. If you can isolate the most important variables you could perhaps create some guidelines, for example a size limitation for database storage.

If you for example have a load balanced application then the database is a very convenient place for several servers to store data in the same place. To get the same function for a different storage you need a way to access the storage and a way to synchronise the access. This means more work for the developers, so it comes down to a cost for software development and maintainence compared to a cost for database setup and maintainence.

Guffa
  • 687,336
  • 108
  • 737
  • 1,005