CODEWITHSUNDEEP
javasqldatabase
Paul Manjarres
Paul Manjarres

Reputation: 48

Storing files on Database

i have to save a file in any format (XLS, PDF, DOC, JPG ....) in a database using Java. in my experience i would have do this by storing the binary data of the file into a BLOB type field, someone told me that an alternative is coding the binary data as Text using BASE64 and store the string in a TEXT type field. Which one is the best option to performn this task?.

Thanks.

Paul Manjarres

Upvotes: 2

Views: 490

Answers (4)

user2102266
user2102266

Reputation: 539

  • FILE to BLOB = DB will not query the content and treat it as, as a ... well ... a meaningless binary BLOB regardless of its content. DB knows this field may be 1KB or 1GB and allocates resources accordingly.
  • FILE to TEXT = DB can query this thing. Strings can be searched replaced modified in the file. But this time DBMS will spend more resources to make this thing work. There may be a 100 char long text inside a field which may or may not be storing 1 million char long text. Files can have any kind of text encoding and invalid characters may be lost due to table/DB encoding settings.No need to use this if content of the files will not be used in SQL queries.
  • BASE64 = Converts any content to a lovely super valid text. A work around to bypass every compatibility issue. Store anywhere, print it, telegraph it, write it on a paper, convert your favorite selfie to a private key. Output will be meaningless and bigger but it will be an ordinary text.

Upvotes: 0

Confucuis
Confucuis

Reputation: 1

Use BLOB to put them in database

Upvotes: 0

Robert Christian
Robert Christian

Reputation: 18310

The argument for using BLOB is that it takes fewer CPU cycles, less disk and network i/o, less code, and reduces the likelihood of bugs:

  1. As Will Hartung says, using BLOB enables you to skip the encode/decode steps, which will reduce CPU cycles. Moreover, there are many Java libraries for Base64 encoding and decoding, and there are nuances in implementation (ie PEM line wraps). This means to be safe, the same library should be used for encoding and decoding. This creates an unnecessary coupling between the application which creates the record, and the application that reads the record.
  2. The encoded output will be larger than the raw bytes, which means it will take up more disk space (and network i/o).

Upvotes: 1

Will Hartung
Will Hartung

Reputation: 118641

BLOB would be better, simply because you can use a byte[] data type, and you don't have to encode/decode from BASE64. No reason to use BASE64 for simple storage.

Upvotes: 8

Related Questions