1 / 21

Materials Microcharacterization Collaboratory

Materials Microcharacterization Collaboratory. Security and Instrument Safety James A. Rome ORNL. Aspects of security. Site protection Strong user authentication Fine-grained authorization Data integrity Disclosure protection via encryption Instrumentation control protocols

payers
Download Presentation

Materials Microcharacterization Collaboratory

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Materials Microcharacterization Collaboratory Security and Instrument Safety James A. Rome ORNL

  2. Aspects of security • Site protection • Strong user authentication • Fine-grained authorization • Data integrity • Disclosure protection via encryption • Instrumentation control protocols • Inter-site communication

  3. MMC security challenges • Diversity of platforms at facilities and at users • Broad, diverse user group • Some proprietary information • Inability to require users to install lots of security HW/SW, especially if it isn’t free • Multiple security venues: • Online instrument operation and control • Data analysis and archiving • Video conferencing

  4. Basic site protection • We think that most resources should be behind a firewall provided that it is • Transparent enough to not “get in the way” • Fast enough to handle the throughput • Cheap enough to be affordable • The Checkpoint Firewall-1 meets these requirements • $2995 for 25 users (behind the firewall) • 40 Mbps throughput

  5. Harmonization • Because of the diversity of our sites and users, we propose to use Web-based remote access and Web-based remote applications wherever possible. • SSL provides encryption • Small user learning curve • Coming ability to use client public key as the basis for all user interactions • Many of the MMC facilities are already online with a large base of users.

  6. Quick and dirty solutions • We need to get things up and running ASAP because the “nice” solutions will take some time to implement. • Several sites use Timbuktu (encrypted tunnels) • General control of the stage and focus of a microscope are straightforward and can be harmonized behind a Web interface • Site-specific features may have to be “out of band” for a while

  7. Scale of the collaboratory • The scalability of security solutions is always an issue. • The MMC will have no more than a few hundred users • Can handle certificates and authorizations manually if necessary • The researchers are generally “trustworthy” folks • No need for big revocation lists

  8. Authentication • Many excellent authentication schemes exist, but most are not available on all platforms • Smart cards and tokens • Kerberos and ssh • X.509 certificates • Biometrics • One-time passwords (S-key)

  9. MMC authentication • Our solution is to use SSL client certificates • This public key is his identity for the MMC • The MMC will issue and sign the certificates • Entrust WebCA handles this for $1/certificate • Downloaded to the user’s web browser online • In Netscape 4.0 these certificates and the corresponding keys can be extracted and used for other purposes • S/MIME for secure E-mail

  10. Authentication conclusions • The client certificate scheme has numerous advantages • Platform independent • Cheap • User friendly — not even a uid/pwd to type • Can be used as the basis for other authorization • But, • The user must protect his keys and Browser

  11. Authorization • Traditionally, enforced by file access restrictions. • File access controls alone are not flexible enough for the MMC • File access controls may be good enough for protecting data • Fine-grained authorizations require authorization certificates

  12. Authorization scenario • To use an online facility, we need proof that • The user has received (and passed) training • A reservation has been made for a time slot • A payment may be required • Additional information is required about the user • Is the work proprietary? • Is the user a student, staff, or researcher? • What site resources does the user need? • These have nothing to do with file access controls

  13. SPKI Certificates • Rather than binding a public key to an identity, what is really wanted is to bind a public key to an action or authorization. This is the goal of SPKI (simple public key infrastructure). • http://www.clark.net/pub/cme/spki-reqts.html • http://www.clark.net/pub/cme/html/spki.txt • http://theory.lcs.mit.edu/~rivest/publications.html

  14. SPKI certificates have 5 parts • ISSUER: The public key of the issuing party both as a name for the issuer and a means to verify the certificate • SUBJECT: The public key receiving authority via this certificate • AUTHORITY: The specific authorization(s) delegated by this certificate (may be delegated to another subject) • VALIDITY: At least an expiration date, but perhaps also a means of online verification (such as a URL) • SIGNATURE: Signature of the issuer (and optionally) the subject to accept the authority granted) • “<issuer> says that <subject> has attribute <auth>”

  15. SPKI trust model • If a verifier is principal “Self”, then the only 5-tuple he or she can trust is of the form • <Self, X, *, A, V> • where • X is the subject asking to be trusted • A is the authority to be granted • V includes the present time • I.e., you are the only authority you can really trust to issue a certificate.

  16. 5-tuple reduction • Ignoring the signature field, a SPKI certificate can be represented as a 5-tuple: • <Issuer, Subject, Delegation, Authority, Validity> • I can delegate the issuing authority to you: • <me, you, D1, A1, V1> + <you, your_user, D2, A2, V2> = • <me, your_user, 0, A, V> • where D1 >D2 A = intersection (A1,A2) V = intersection (V1,V2)

  17. PolicyMaker • Sometimes, credentials don’t grant the exact <auth> needed, A. Instead, one has a policy which, in effect, accepts <auth>s A1,A2,A3 to be used instead of A. • PolicyMaker (ftp://research.att.com/dist/mab/policymaker.ps)allows the formulation of these more complicated (non-intersection) policies. • Example: you might need authorization from 3 out of 5 Vice Presidents to obtain authorization for a check of $300,000.

  18. MMC authorization certificates • We propose to use SPKI certificates to instantiate the bindings between a user’s public key (from his browser client certificate) and each authorization. • LBNL (Larsen has agreed to provide a certificate engine for us to use by the end of this FY) • We propose to store the certificates as Cookies on the user’s Web browser • We will create policy engines to combine multiple input certificates into single device certificates

  19. Implementation — Year One • Put sites behind firewall to stop most Internet-based attacks • Implement and issue the SSL2 Client certificates • Survey the sites to determine their needs • Implement secure Web servers to provide encrypted access to researcher’s E-notebooks • Obtain authority certificates from LBNL

  20. Implementation — Year Two • Required SPKI certificates must be determined and created • A certificate acquisition process must be created and implemented • What certificates does a user need? • Where are they obtained? • PolicyMaker engines created, integrated, tested • Pilot deployment at a few sites

  21. Implementation — Year Three • Deploy the infrastructure across the MMC • Provide cross-realm delegation (if desired) • Implement SPKI security for data analysis tools • Fix problems as they arise

More Related