Over the past couple of years quite a few folks have asked me, and I’m sure others, “what’s the salient differences between OpenID and SAML?” So earlier this year I began hacking together a technical comparison of the two. It’s an interesting exercise comparing two Web SSO protocols, even one as ostensibly simple, and straightforwardly specified, as OpenID. It turns out to be a fairly complex task given all the different facets inherent in authentication protocols in general, and in web-, i.e. HTTP-based, protocols (and profiles thereof) in particular. And also given the various audiences affected by such protocols: implementors, deployers, end users, and protocol designers.
The resultant comparison paper, “Technical Comparison: OpenID and SAML – Draft 05” seems to me to be at a stage where it can be shared widely (i.e. on the web ), here it is..
..For many readers, sections 1, 2, and perhaps 3 ought to cover things. For those necessarily interested in gory, really geeky details, parts or all of section 4 will be of interest. Note that this is still a “draft“–there are various items, especially in section 4, that are not as yet evaluated as thoroughly as I’d like, or at all (as yet).
I’ve tried as much as possible to provide an objective comparison. It’s admittedly difficult given I’ve been intimately involved in SAML’s gestation since essentially the very beginning. It’s also a technically difficult comparison because of the differing design centers of OpenID and SAML, as well as differing specification styles, and thus the difficulty in presenting the comparison to the reader, not to mention attempting to be “balanced“.
So, I hope this paper will prove at least somewhat enlightening and useful to the multifaceted “identity” community out there, and to those shepherding websites who are wondering what these two oft-mentioned beasts are, how’re they’re different/similar/alike, and also nominally how they work.
=JeffH sez check it out.