What is it exactly a BLOB in a DBMS context

BLOB :

BLOB (Binary Large Object) is a large object data type in the database system. BLOB could store a large chunk of data, document types and even media files like audio or video files. BLOB fields allocate space only whenever the content in the field is utilized. BLOB allocates spaces in Giga Bytes.

USAGE OF BLOB :

You can write a binary large object (BLOB) to a database as either binary or character data, depending on the type of field at your data source. To write a BLOB value to your database, issue the appropriate INSERT or UPDATE statement and pass the BLOB value as an input parameter. If your BLOB is stored as text, such as a SQL Server text field, you can pass the BLOB as a string parameter. If the BLOB is stored in binary format, such as a SQL Server image field, you can pass an array of type byte as a binary parameter.

A useful link : Storing documents as BLOB in Database - Any disadvantages ?

I won't expand the acronym yet again... but I will add some nuance to the other definition: you can store any data in a blob regardless of other byte interpretations they may have. Text can be stored in a blob, but you would be better off with a CLOB if you have that option.

There should be no differences between BLOBS across databases in the sense that after you have saved and retrieved the data it is unchanged.... how each database achieves that is a blackbox and thankfully almost without exception irrelevant. The manner of interacting with BLOBS, however can be very different since there are no specifications in SQL standards (or standards in the specifications?) for it. Usually you will have to invoke procedures/functions to save retrieve them, and limiting any query based on the contents of a BLOB is nearly impossible if not prohibited.

Among the other stuff enumerated as binary data, you can also store binary representations of text -> character codes with a given encoding... without actually knowing or specifying the encoding used.

BLOBS are the lowest common denominators of storage formats.

This may seem like a silly question, but what do you actually want to use a RDBMS for ?

If you just want to store files, then the operating system's filesystem is generally adequate. An RDBMS is generally used for structured data and (except for embedded ones like SQLite) handling concurrent manipulation of that data (locking etc). Other useful features are security (handling access to the data) and backup/recovery. In the latter, the primary advantage over a regular filesystem backup is being able to recover to a point in time between backups by applying some form of log files.

BLOBs are, as far as the database concerned, unstructured and opaque. Oracle does have some specific ORDSYS types for multi-media objects (eg images) that also have a bunch of metadata attached, and have associated methods (eg rescaling or recolouring an image).