• Open Menu Close Menu
  • Apple
  • Shopping Bag
  • Apple
  • Mac
  • iPad
  • iPhone
  • Watch
  • TV
  • Music
  • Support
  • Search apple.com
  • Shopping Bag

Lists

Open Menu Close Menu
  • Terms and Conditions
  • Lists hosted on this site
  • Email the Postmaster
  • Tips for posting to public mailing lists
Re: Xcode... - rendezvous + cluster
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Xcode... - rendezvous + cluster


  • Subject: Re: Xcode... - rendezvous + cluster
  • From: Wade Tregaskis <email@hidden>
  • Date: Wed, 25 Jun 2003 13:30:16 +1000

What you've described is effectively what's being pushed as 'grid computing'. It is not a part of the OS - it is a whole new world in which to find exotic problems. Not the least of which is security - would you trust anyone with network access to your machine to run code on it?

On the local link? Generally, yes. Making Rendezvous link-local only turns a network security problem into a site security problem, and those are much easier to solve.

You can never trust your network. Networks simply don't provide much security against unwanted access . Maybe if you're using a VPN or Airport, but even then there are still issues...

Plus it may be that you don't trust everyone on your network, and for good reason. I'm on a network here at college with at least 200 computers on my subnet, and I don't think I'd trust a single one of them implicitly. What I might do is trust some of them with certain limitations, like not letting them use any networking accept back to the original computer (to prevent my computer becoming a willing part of a DDoS, for example), and not having access to local storage except in a particular directory (e.g. /tmp/<someuniquedirjustforthatprocess>/), etc. There's no reliable mechanism in MacOS X to do this - if systrace were integrated, then we'd have something to work with.

The purpose of my aforementioned project is to overcome these problems and others - like user authentication. MacOS X doesn't have any scheme for distributed PKI operations. My current subproject is a 'Keychain' framework to provide just such a unified scheme, but it's at 20,000 lines and growing with still only very basic functionality. It was meant to just be an ObjC wrapper over all Apple's C/C++ stuff, but due to the limitations of that stuff*, I'm having to write some functionality from scratch. Not fun.

So even on a local network (which I wouldn't assume; what if you have a server farm of a thousand machines, on multiple subnets, at a remote location?) there are a lot of issues which need to be solved, and don't have much OS-level support [as yet]. Having said that, the Security guys seem to be full of secrets lately, so I'm somewhat hopeful that there'll be some new stuff in Panther (perhaps someone could check the Security framework in the preview release for me), but that's up to 6 months away. I'd originally planned to finish this project within 6 months from now. :/

Wade Tregaskis
-- Sed quis custodiet ipsos custodes?

* = This isn't having a dig at the Apple guys - they've done a heck of a lot of good work all round, and kudo's to them for it. Unfortunately for me and them, however, there's still a lot to be done.
_______________________________________________
cocoa-dev mailing list | email@hidden
Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/cocoa-dev
Do not post admin requests to the list. They will be ignored.
References: 
 >Re: Xcode... - rendezvous + cluster (From: Jeff Harrell <email@hidden>)

  • Prev by Date: Re: Safari SDK
  • Next by Date: Re: Mouse tracking an NSCell in my control
  • Previous by thread: Re: Xcode... - rendezvous + cluster
  • Next by thread: WebFrame does not respond to loadRequest: ?
  • Index(es):
    • Date
    • Thread