首先，你应该期望 ，经常显着地）较慢，但绝对不比将它们存储在文件系统中快。将它们存储在DB中的原因不以性能为中心，而是关于例如：

First of all, you should expect storing BLOBs in a database to be (sometimes a bit, often significantly) slower, but definitly not faster than storing them in a file system. The reasons to store them in a DB do not center about performance, but about e.g.:

• 在（共享）文件系统的不可用性群集或负载平衡的场景


• 易于备份：单个进程，aot 2进程使用文件和数据库时


• 事务安全：BLOB已完成或尚未完成，但尚未完成。


• 其他我现在不能想到的。

• Unavailability of a (shared) file system in a clustered or load-balanced scenario
• Ease of backup: Single process, a.o.t. 2 processes when files and DB are used
• Transaction safety: A BLOB is either there and complete or not, but not in a half-baked stage
• others I can't think of right now.

一般的经验法则是，如果这些都不关心你，你应将您的文件存储为...文件。将元数据和路径名存储在DB中是IMHO好的和常见的做法。

The general rule of thumb is, that if none of these concern you, you should store your files as ... files. Storing the metadata and pathname in a DB is IMHO good and common practice.

关于Oracle调整：有关于这些的书。我怀疑他们总计超过一吨的死树平装书格式。你可能首先看看Oracle进程的内存消耗 - 经验法则：如果它不到一个gig，你使用BLOB，你就麻烦了。在不同的内存池上读取以及如何增加它们。对于快递版本的某些限制可能适用。

Concerning Oracle tuning: There are books written about that. I suspect to total them far over a ton in dead-tree-paperback format. You might first of all look at the Oracle process' memory consumption - rule of thumb: If it is less than a gig and you use BLOBs, you are in trouble. Read up on the different memory pools and how to increase them. Some limits for the express edition might apply.