[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Last Call & idea for OCSP
Stef--
Work on OCSP is proceeding. A draft incorporating comments to date should
be available for final review in about two weeks.
Regarding an incremental certificate validity list, a server could equally
well cache prior signed responses per certificate. Sending back just that
response vs. a list of other valid certs that are potentially relevant
saves bandwidth. One might also consider the use of the HTTP "Expires:"
header field to yield cacheing effects on the requestor's Web server.
--Mike
At 03:55 PM 10/1/97 +0200, you wrote:
>
>And the OCSP protocol?
>
>This may be an idea for that protocol, so the CA doesn't have
>to sign each answer separately (which takes quite some time):
>
>- The CA divides its certs in groups of say 100 or 500
>- When an OCSP_query for a cert C comes in:
> * If the cert is not valid (revoked, expired, ...)
> the CA replies with an ordinary OCSP-response.
> * If the cert is valid, the CA makes a list of
> all valid certs that are in the same group as cert C.
> The CA then signs this list together with the current
> time and sends it to the requestor. That one must look
> for his cert C in the list to assure C is valid.
>
>If another OSCP-query for a valid cert in the same group
>as C comes in, within say a minute, the CA can simply
>send the same message as for cert C without having to
>sign anything.
>
>(This ressembles a bit the CRL functionality but CRL's
>may be much larger then this here.)
>
>(It's all based on Micalli's paper on the previous
>RSA conference.)
>
>
>Don't know if it's no old news or not realistic ...
>
>Stef
>
>-------------------------------------------------------------------------
>stefan.hoeben@esat.kuleuven.ac.be | All wiyht.
>Katholieke Universiteit Leuven | Rho sritched
>Dept. Electrical Engineering-ESAT / COSIC | mg kegtops
>K. Mercierlaan 94, B-3001 Heverlee, BELGIUM | awound?
>tel. +32 16 32 10 73 fax. +32 16 32 19 86 |
>-------------------------------------------------------------------------
>
>
>