|
Re: x.509 v3 Certificates and Compatbility: msg#00285ietf.x509
Stefan, I would recommend use of the Certificate Policies extension to further assist in selecting appropriate certificates. A critical Certificate Policies extension should ensure that the certificate is used only for the purpose for which it is intended. Dave S. Stefan Santesson wrote: > > All, > > I agree with Bob. > > We have come to the same conclusion. There may be several reasons > for having more than 1 certificate even for the same key usage. > > More reasons may be: > - Different certificate policies > - Different integrity aspects (some personal information should not be > published while some shouldn't) > - Different validity periods for different personal data. > - Some attributes can't be certified by the same CA:s. (i.e. Bank account > and employment number) > etc. etc. > > Compare with, gasoline cards, ID-cards, credit cards, library cards, > employment cards, medical cards. They are all "physical" certificates > certifying my connection with different attributes and privileges. > > My concern is mainly. How do the certificate holder select the appropriate > certificate. > Suppose that the entity has two certificates with the same key usage. One > anonymous for his www.sex.com and one digital ID certificate for banking > applications over the internet. In both cases the applications is run over > http. > > Will there be any suitable mechanisms that selects the appropriate > certificate. Is there any actions that can be taken by the server to help > the client to select the appropriate certificate or will the entity be > forced to select by him self? > > Is there any one with a sound solution to this problem? Any opinion? > I would like to se a solution that doesn't require extra, non standard, > modules for the client. > > /Stefan > > At 02:11 PM 8/17/98 -0600, Bob Jueneman wrote: > >Brad, > > > >Your question is quite reasonable. In fact, I went through such > >an analsysis with respect to our own PKI product just last week. > > > >Michael Smith has provided one answer, but consider the following: > > > >1. You may need to correspond with various people throughout > >the world, and they with you. Some of those people (a steadily > >decreasing number, it seems) may be unrestricted in terms of what kind of > >crypto they have available to them. In order to support those people > >you may need to offer several different kinds of key lengths that > >people can choose between, including 2048 bit, 1024 bit, > >and 512 bit. > > > >2. Some users, vendors, and/or standards groups may have a particular > >bias for or against some particular algorithm, perhaps because of patent > >considerations, technical strength issues, national pride, time/space > >tradeoffs, etc. As a result, you may need to support RSA, DSS, elliptic > curves, > >Skipjack/KEA, Red Pike, GOST, etc., etc., together with all of the various > >key length variations. > > > >3. Some countries won't allow the use of encryption unless the traffic in > >both directions is accessible to the governmen authorities through some > >form of key escrow. Therefore, even if your government doesn't require > >key escrow, you may be forced to use it to communicate securely (sic) > >with someone in France, for example. So you may have to escrow a set > >of keys for that prupose, for example, but you would probably not want to > >use those keys for any other purpose. > > > >4. For export/import/usage controls reasons, it is desirable to > >separate the use of encryption and digital signature keys, since digital > >signature (only) keys generally receive a more favorable export treatment. > > > >5. The digital signature vs. nonrepudiation key usage argument is > voluminous. > >Let me just say that it wouldn't be a good idea to use the same key pair for > >authentication (say to your favorite www.sex.com web site) as is used > >for signing an important legal document. > > > >6. There are key roll-over and other issues that will have to be dealt with. > > > >7. You may have different roles -- personal, employee, secretary of the > >bowling league, etc. > > > >Bottom line: considering only RSA, DSS, and ECC for signing, plus RSA > >and D-H for encryption, plus a couple more for mixed usage a la SSL, times > >three for different key lengths, and not including any role-based > authentication > >or key escrow considerations, it would not surprise me to see someone with > >on the order of 20 certificates. Now throw in the restricted use > certificates, > >such as those used for SET and perhaps issued by different banks, > certificates > >containing proprietary attributes that may be used by your company, etc., > >and you'll bein to see why I doubt that they are all going to fit on a > single smart card! > > > >BTW, this is a very real consideration when designing directory storage > >and access mechanisms. > > > >Bob > > > > > >Robert R. Jueneman > >Security Architect > >Novell, Inc. > >Network Products Group > >122 East 1700 South > >Provo, UT 84604 > >801/861-7387 > >bjueneman@xxxxxxxxxx > > > >"Architects, like prophets and saints, are > >usually years ahead of their time. For this > >reason they are often difficult to live with > >at home." > > > > > >>>> Mike Smith <mfsmith@xxxxxxxxxxxxx> 08/13 4:54 PM >>> > >Domains of trust. If your single cert came from a single source that > EVERYONE (worldwide) trusted (and who indemnified ALL who relied on the > certs they issued and that their authentication practices met or exceeded > those in other domains of trust) and they issued all the rights to you at > once, then, maybe that single cert could be practical. However, I'm still > not sure I would trust it for anything other than issuing a cert to you to > do business from me. > > > >michael > >>>> brad h <bradh_1998@xxxxxxxxx> 08/13 2:46 PM >>> > >I have a question that the group might be able to help me out with. > >I've been researching this question but have not yet been able to come > >up with an answer. > > > >I was wondering why a person would have to have more than one x.509 v3 > >certificate? From what I understand they should all be inter-operable. > > > >If you have a x.509 v3 cert shouldn't you be able to add extensions > >for each type of device/solution that you're trying to access (ex. > >SSL, S/Mime, VPN, PKI, etc.)? > > > >Brad > > > > > > > > > > > >_________________________________________________________ > >DO YOU YAHOO!? > >Get your free @yahoo.com address at http://mail.yahoo.com > > > > > > ! > >! > >! > > > > > > > > > ------------------------------------------------------------------- > Stefan Santesson <stefan@xxxxxxxxxxx> > Accurata Systemsäkerhet AB > Lotsgatan 27 D Tel. +46-40 152211 > 216 42 Malmö Fax. +46-40 150790 > Sweden Mobile +46-70 5247799 > > PGP fingerprint: 89BC 6C79 5B3D 591B 8547 1512 7D11 DBF4 528F 29A0 > ------------------------------------------------------------------- -- David Simonetti, Booz·Allen & Hamilton Inc. |
|
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| Previous by Date: | Re: block padding formats: 00285, Russ Housley |
|---|---|
| Next by Date: | PKIX LDAPv3 Schema questions: 00285, Anne Anderson - Sun Microsystems |
| Previous by Thread: | Re: x.509 v3 Certificates and Compatbilityi: 00285, Patrik Nilsson |
| Next by Thread: | Re: x.509 v3 Certificates and Compatbility: 00285, Stefan Santesson |
| Indexes: | [Date] [Thread] [Top] [All Lists] |
| News | FAQ | advertise |