We have a CA that has thousands of revoked certificates which
leads to CRLs os several MBytes.
On the next nenewal of the CA, we are thinking of partitioning the
CRLs at each X number of issued certificates. The issued certificates
will have different CRL Distribution Points (CDP) according to the
partitions they are assigned.
For example, for X=100, from certificate 1 to certificate 100, the
CDP would be http://myca.com/crl/myca-0001.crl, from certificate 101
to 200 the CDP would be http://myca.com/crl/myca-0002.crl, and so on.
My question: Are Internet Explorer and IIS prepared to support
partitioned CRLs like the way described? In particular, if CRLs are
cached, they must be able to merge several different partitions
according to the CDP to create a unified view over the revocation
universe of a CA.
Regards,
Nuno Ponte
MYCA.crl
MYCA(1).crl
MYCA(2).crl
You could choose to renew at regular intervals, or in your case, renew for
each X certificates issued.
This will add management overhead though, as all *active* CRLs (still time
valid) must be published to the CDP locations.
Brian
"Nuno Ponte" <nuno....@gmail.com> wrote in message
news:c786e7cf-1267-4f91...@s20g2000prd.googlegroups.com...
Thanks for your reply.
The CA is not a Microsoft CA. It's an open source CA, so we can
develop new funcionalities.
My question is about the ability for IIS and IE to support such a
partitioning scheme at validation time of a certificate. I am trying
to figure out if there is support from the applications before
starting to implement it.
Regards,
Nuno
> The CA is not a Microsoft CA. It's an open source CA, so we can
> develop new funcionalities.
> My question is about the ability for IIS and IE to support such a
> partitioning scheme at validation time of a certificate. I am trying
> to figure out if there is support from the applications before
> starting to implement it.
Both of those applications use the crypto functions built into the OS, and
the crypto functions built into the OS are RFC compliant. As long as what
you're trying to do is also RFC compliant then there shouldn't be a
problem.
--
Paul Adare
MVP - Identity Lifecycle Manager
http://www.identit.ca
The problem is that sometimes standards are not clear enough, leading
to ad-hoc interpretations (and implementations). :-(
For example, in RFC 3280, the definition for the CRL "nextUpate" field
does not clearly state it is an expiricy date (such as the notAfter
field on certificates), but most of the people interpret (and
implement) it in fact as an expiricy date (http://www.imc.org/ietf-
pkix/mail-archive/msg03166.html).
> The problem is that sometimes standards are not clear enough, leading
> to ad-hoc interpretations (and implementations). :-(
Then I think at this point you're either going to have to open a support
case with Microsoft or simply test this yourself.