couchdb vs mysql速度

Question

What can you tell about speed of couchdb and mysql databases?

I mean, very simple requests like getting one row (or one document) by unique id and simple requests like getting the 20 ids/rows/documents with the biggest date (of course, using indexes and views etc - don't really know how it works in CouchDB but I'm pretty sure there is something. Please don't send me learn how CouchDB works: I'm going to learn it, but I need a perfomance comparison anyway).

Thanks!

---

As I realized from the links in the first answer, while I have only one server for DB, it's a lot better to use MySQL?

Answer 1

我发现这篇文章与另一个关于MySQL与CouchDB性能的问题相关联，但我不知道它是否重要。

Answer 2

Firstly, note that CouchDB is not a relational SQL database like MySQL. I am not sure whether the query you describe can be expressed in CouchDB (but I'm no expert on this).

That being said, here are some links:

• http://metalelf0dev.blogspot.com/2008/09/mysql-couchdb-performance-comparison.html
• http://www.codeweblog.com/couchdb-vs-mysql-insert-performance-test-data-of-the-speed-test/

• When to use CouchDB vs RDBMS

• https://saucelabs.com/blog/goodbye-couchdb

Answer 3

One way that couchdb could facilitate a dramatic speed improvement, is with its replication. It can replicate across high latency connections, perhaps even down to a client site if the security infrastructure supports it. This way you could have multiple databases at multiple physical locations without killing yourself in complexity.

When I did some comparisons on my own, I found couch to be about half the speed of oracle for a scenario with high numbers of inserts 100k inserts while readers were reading. This of course is not couch's strength, as the readers had to rebuild the views while inserts were occurring.

Answer 4

I would recommend that you set up a test yourself. Why ? CouchDb and MySql are very different beasts, and how you would choose to organise data in each would be very dependent on your problem domain. Consequently how you query them will differ as well.

I think that some dummy data in each and some benchmarking by yourself will be more worthwhile than looking at the results of some arbitrary test. I realise this will take some time in itself, but I think it's the best way forward.

Answer 5

Exactly. The main diference is that: Relational vs Non Relational. And it depends of what you need. There is a movement about this: http://en.wikipedia.org/wiki/NoSQL

And, another good no-relational db:

www.mongodb.org

Answer 6

我听过的最好的比喻之一是CouchDB不想成为法拉利，它希望成为本田，即不是最快，但最可靠。