Is it possible to retrieve a specific page from Akamai's cache?

[thorin]

Ok so I have a hypothetical to bounce off everyone.

Say I know that my client ABC uses akamai, with the following hypothetical/redacted DNS details:

host www.ABC.com
www.ABC.com is an alias for www.ABC.com.edgekey.net.
www.ABC.com.edgekey.net is an alias for e1111.b.akamaiedge.net.
e1111.b.akamaiedge.net has address 2x.x.x.16

It's my understanding that e1111 makes 1111 the Akamai customer number (based on the out-dated MS Research paper below).

Does anyone know of a way to request a specific URI from Akamai so that I can see what they have cached?

I found this write-up but it seems out of date:
http://research.microsoft.com/en-us/...ul/akamai.html

[hhmatt]

I haven't tried or tested any of this but at first glance I don't see why you couldn't. If that paper isn't completely out of date.

I don't know if you missed this but this seems to be the most important part here:

* update (courtesy John Jacob): The hex string is created using "md5sum [file_name] | cut -c3-16". It can also be replaced by a cache time-to -live value like "1d" for one day, and "15m" for 15 mins.

So the only problem you face is knowing the client id, which you seem to already have, and the other seemingly random number that comes right before the hex string. I would think you could brute force that number, it seems to fall within a 1024 or less number range.

Why would you specifically need akamai's cache wouldn't there be others that you could use?

Anyways, hope this helped and good luck!

[thorin]

Thanks hhmatt, I did try doing TTL values (i.e. 1d etc as suggested) but I didn't actually do the md5sum, I'll try that. What's sadly unclear is whether [file_name] is the final part of the URL or the entire path... I'll have to see where trial and error gets me.

[hhmatt]

I found this to be pretty interesting, thanks for bringing this up!

I figured they would probably patch the TTL insertion or maybe it's in a different form like hex. It maybe possible to send some malformed http requests that may trick it into believing the md5sum is true. I would also give 0 and 1 a shot.

Getting a md5sum of the file you are trying to receive is troublesome.

[thorin]

Just in case anyone else is interested I'm still trying to figure this out. I haven't had any luck yet even using md5sums.

If you have any brain waves please let me know. If I come up with anything useful I'll update this thread.