?

Log in

No account? Create an account

Webdav vs S3

« previous entry | next entry »
May. 29th, 2007 | 12:20 pm

Putting aside the infrastructure that Amazon has put into place for
S3 Service, think about the S3 from an API standpoint vs Webdav.

Webdav is an HTTP protocol for sharing files. It makes an http server
appear as though it is a local filesystem. For webdav the HTTP
protocol was extended for PROPFIND, PROPPATCH, MKCOL, COPY, MOVE,
LOCK and UNLOCK. It also makes use of the DELETE and PUT keywords. S3
is a bit different, it takes a REST approach and uses just GET, PUT,
and DELETE.

Its simple, straight forward, and very easy to code too.

S3 though is not a protocol, it is a service.

S3 as a popular service creates a de-facto API.

Now does S3 do everything that webdav provides? No, but it does the
core pieces well.

So what I am left wondering is, has webdav been relegated to being
BETA while S3 is VHS?

For the vendors who will spring up to compete with Amazon, will they
copy the S3 protocol, or will they go with Webdav?

What does this mean for a database like MySQL, or do we want to keep with the mindset that we will always have local disk?

Link | Leave a comment | Share

Comments {4}

Jamz

(no subject)

from: woggie
date: May. 29th, 2007 10:51 pm (UTC)
Link

To me, S3 is a brand of monitor and therefore video driver. From that standpoint, I'm not sure what you've said makes sense.

Reply | Thread

(no subject)

from: georgekodinov
date: May. 30th, 2007 07:58 am (UTC)
Link

There's more to WebDAV than just the filesystem analogy : it's Web Distributed Authoring *and Versioning*.
The versioning part is the one that is still under consideration afair.
Besides : WebDAV (sans versioning) is implemented in all the major OSes (WinXP and later, MacOSX and Linux) at a file system level. So it's just an alternative of SMB or NFS for example, not something the vendors should interface with IMHO (except if they want the extra benefit of the speed, but I'm not sure there are a lot of products these days that integrate to SMB for example).

Reply | Thread

Brian "Krow" Aker

(no subject)

from: krow
date: May. 30th, 2007 04:52 pm (UTC)
Link

I acknowledged that Webdav can do quite a bit more then S3, versioning being one of those things.

Whether something is at the OS level, aka file share, or not is pretty immaterial to me. Everything is a storage API, the question to me is how useful is it in a particular environment. The open()/close() everything as a file UNIX method is not likely to go away anytime soon... really Fuse is the first piece of generic work to really extend it.

Reply | Parent | Thread

cudddly

(no subject)

from: cudddly
date: May. 30th, 2007 01:05 pm (UTC)
Link

WebDAV is a particularly hateful protocol. There's a nice spec and a spec tester (litmus) but all the clients completely ignore the spec. There are various things that provide WebDAV for your S3 account like JungleDisk. Using WebDAV over high latency is very annoying, so the local WebDAV server which does S3 in the background works well. I'm bitter, I never managed to get Net::DAV::Server to be good enough...

Reply | Thread