哪个数据库空间效率最高
目前我们使用Oracle来存储日志信息。其中一列是存储 XML 负载的 blob。我们每周都会归档日志模式，但我们仍然会遇到 > > 的情况。每周 1TB。
目标：减少数据库的空间占用
解决方案1： 在将 XML 有效负载放入 BLOB 之前对其进行压缩。
解决方案2： 查找已默认压缩数据的数据库（noSQL 或其他关系数据库）。 JDBC 驱动程序必须可用。
有人知道适合解决方案 2 的数据库吗？这样我们就不需要修改应用程序，只需要更改数据库并更新 JDBC 驱动程序即可。
Current we are using Oracle to store logging information. One of the column is a blob that stores an XML payload. We archive the logging schema every week, but we still hit about > 1TB a week.
Goal: Reduce space footprint of database
Solution 1:
Compress the XML payload before putting it into the BLOB.
Solution 2:
Look for a database (noSQL or other relational DB) that already compresses data on default. JDBC driver must be available.
Anyone know a database that fits into solution 2? So that we do not need to modify the app, we just need to change the DB and update the JDBC drivers.
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（3）
Oracle 支持几种不同级别的压缩 - 可以轻松地适应解决方案 2。 （表/行/lob 压缩）
Oracle supports a few different levels of compression - that fits in to solution 2 with minimal effort. (Table / row / lob compression)
据我所知，所有 dbms 都支持压缩和 JDBC。
PostgreSQL 依赖底层操作系统进行存储。您可以在压缩文件系统上构建 PostgreSQL 表空间。
All the dbms that I know of that support both compression and JDBC.
PostgreSQL relies on the underlying operating system for storage. You could probably build PostgreSQL tablespaces on a compressed filesystem.
首先，我想说，如果您有企业，则对该 xml 表进行分区并定期备份和修剪它（如果按日期范围分区，则删除早于 x 的分区）。如果这不是一个选择，您可以尝试 noSQL 数据存储。
有很多可用的 noSQL 选项。我玩过的一个（尚未生产）是 mongoDB。它以二进制 JSON 格式 (BSON) 存储数据，并且可以压缩。 更大的公司似乎也取得了进展。前段时间我曾经研究过一个类似的b-tree平面文件系统，速度非常快。但您不会拥有与 RDBMS 相同的功能。 mongoDB 的好处是它由 10gen 提供商业支持（同样，不能说这种支持有多好，但这对某些商店来说很重要）。
这是一篇很好的文章，讨论了更大的 noSQL 的优缺点那里有选择。
不，我不为 10gen 工作；）当然，要做更多研究并决定什么最适合您的需求
First, I would say if you have enterprise, then partition that xml table and periodically backup and prune it (if partitioned by date range, then remove partitions older than x). If that isn't an option, you may try a noSQL data store.
There are plenty of noSQL options available. One I've played with (not yet production) is mongoDB. It stores the data in a binary JSON format (BSON), and can be compressed. It seems to be making headway with bigger companies as well. I once worked on a similar b-tree flat file system some time ago, was very fast. You won't have the same features as a RDBMS though. Good thing about mongoDB is that its commercially supported by 10gen (again, can't say how good that support is, but that's important to some shops).
Here's a good article discussing the pros/cons of the bigger noSQL options out there.
And no, I don't work for 10gen ;) Of course do more research and decide what's best for your needs