Skip to content
Snippets Groups Projects

Performance with git

Closed Marc Radulescu requested to merge performance_with_git into master
1 file
+ 5
4
Compare changes
  • Side-by-side
  • Inline
@@ -12,13 +12,14 @@ The switch from a centralized version control system such as CVS or SVN to Git i
- speed
- data integrity
In Git, The main repository (origin) lies on the server, and developer machines have local copies. Developers commit to their local machines, and they only need to connect to the server when pushing their work back to the main repo.
In Git, The main repository (origin) lies on the server, and developer machines have local copies.
Developers commit to their local machines, and they only need to connect to the server when pushing their work back to the main repo.
This means that even in the case of a server downtime, development can still continue on the developers' local clones.
**Server downtime is not as disruptive to the workflow as with a centralized tool.**
After the first cloning of the origin repo, Git only registers the changes to the files, so the deltas between the origin and the clones are very small.
That's less information that needs to travel through the wire.
**Getting the clone in sync with the origin feels instantaneous in most of the cases.**
**Getting the clone in sync with the origin does not take a lot of extra time if you are geographically further away.**
Code searching is also considerably faster than with a centralized tool, because the search happens on the local clone of the repo.
@@ -29,6 +30,6 @@ Replicating slows down the GitLab installation.
We also recommend keeping all developers in the company on one GitLab server, instead of splitting it into several smaller servers for different groups/departments.
The rationale for this is that user management becomes more efficient as the GitLab instance is scaled up.
If you are concerned with managing the codebase and keeping it clean, you can consider either [forking](http://blogs.atlassian.com/2013/05/git-branching-and-forking-in-the-enterprise-why-fork/), or using [protected branches](http://doc.gitlab.com/ce/permissions/permissions.html)
If you are concerned with managing the codebase and fine-grained authorizations, you can consider either [forking](http://blogs.atlassian.com/2013/05/git-branching-and-forking-in-the-enterprise-why-fork/), or using [protected branches](http://doc.gitlab.com/ce/permissions/permissions.html)
In case you really feel that replication is needed for your git server, our suggestion is to take a look at products such as the ones put forth by [WANdisco](http://www.wandisco.com/)
In case you do want geogrihic replication for you git server our partner WANdisco offers this in their [GitLab and Git Multisite](http://blogs.wandisco.com/2014/10/20/gitlab-git-multisite-architecture/) product.
Loading