Does Googlebot spider unlinked iframes reliably?

Can you use an IFRAME for site navigation?

[Sharper]

I'm building a new site that it's going to be much simpler for me to return my pages with the main page content in an html page and reference from within that page an IFRAME for the link navigation.

My main concern is that Google will spider the page and treat the IFRAME as an entirely different page, meaning that it will give it seperate pagerank, treat links from it as one-level deeper (assuming it follows the iframe tag as equivelent to a "link"), etc...

Anyone have lots of experience with Googlebot and IFRAMES?

<Begin technical explanation for those who care>

I have a front-end server that will be serving the end-result pages to the world. It is running Apache2 with, among other things, mod_proxy. It will be reverse-proxying the page request to various back-end servers that don't know anything about the directory/navigation layout of the front-end server, where the user is in the site navigation, nor what links should appear on the page.

Thus, I need to get the content using mod_proxy (resulting in an html page from the back-end going to the cliunt) and the navigation from the front-end server that is putting the page together and knows where the user is in the site navigation.

So my current proposed solution is to have the back-end server return it's html including a relative IFRAME URL so that the front-end server can then serve up the file for the IFRAME with the links/navigation that only it knows about.

Of course, if this solution will cause Google to not treat the links in the IFRAME'd page as part of the main page for ranking purposes, that will put a serious cramp in my plans, so I'll need to reevaluate and possibly have to essentially write my own custom proxy so that I can front-end server-side change stuff instead of just using Apache's built-in very useful reverse-proxying features.

It should be pretty obvious how much less work it will be to use the built-in stuff.

<End technical explanation>

Any comments/suggestions will be appreciated. :)

[MHes]

I suspect your navigation will be indexed as a seperate page. You mentioned the pr flow problem and also the links will be on a page with no decent content to back them up. The solution could be to put your navigation into a noscript tag on your main page. However, noscript and noframes etc. has been abused in the past and may be treated with suspicion. I would avoid iframes for navigation. Good luck

[encyclo]

You might want to read this related thread:

[webmasterworld.com...]

Using iframes for navigation are nothing but trouble, even when backed up with a comprehensive site map. Is there any way that your front-end server can build the page with SSI?

[Wizard]

Iframe is good solution if you want to prevent some links from being crawled. If your intentions are opposite, you have to use server-side connecting of the navigation with the content, all other ways will make more complication than benefit.

[Sharper]

So far everyone has just confirmed the problems I thought I might see. That doesn't give me a lot of hope for using an IFRAME.

MHes,
If I could put the links into a no-script tag on the main page from the back-end, I could just as easily just put it in as normal links. :)

Encyclo,
With mod_proxy, SSI isn't available to parse the html file after it comes back from the reverse-proxied back-end server.

Looks like I'll have to go with the Plan B I've since come up with, which is including the navigation reference in the query-string of the URL to the back-end server so that it can take that an fill it into the page. It's a little limiting as there's only so much room in the URL and I'll have to keep a file-map of what links to pass on the query-string for which URL, but I suppose that's what I'll have to do to get around the IFRAME issues.

[MHes]

>MHes,
If I could put the links into a no-script tag on the main page from the back-end, I could just as easily just put it in as normal links. :)

You're right!
.... which just goes to show how stupid I can be :(

Are long url strings still a problem with google? In the old days you needed reasonable pr to get the spider to follow them.