Why is SVN checkout so much slower than CVS?

From the Subversion developer list:
---SNIP---
Svn manipulates lots of small files in the .svn/ area. (It journals all of its actions, for example.) And NTFS is tremendously slow at this, compared to Linux filesystems. For operations that are working-copy intensive, windows seems to be about 2x slower in this regard. There's nothing we can do about it (other than redesign all of the working copy code in svn 2.0.)
---SNIP---

This means that some operations which make a lot of changes to the working copy seem slow. This mainly affects the initial checkout.

On the other hand many operations are much faster. Subversion stores a local copy of the unmodified file, which means that you can diff your local changes without contacting the server at all. You can also revert your changes and start again. And because it has both the new and old versions, subversion only needs to transmit the differences to update and commit, so these operations tend to be much faster over external networks.

Some workarounds for very large projects:
Checkout only a part of your project, not the entire tree.
Run update only on subtrees.
Search for Slow in this FAQ to look for other articles.