archive-org.com » ORG » S » SHUB-INTERNET.ORG

Total: 106

Choose link from "Titles, links and description words view":

Or switch to "Titles and links view".
  • Analysis: Qpopper ìServer Modeî
    111 Notes This does require that incoming mail messages may be temporarily delayed a bit while the user is processing their mailbox but overall I still believe that this is

    Original URL path: http://www.shub-internet.org/brad/papers/dihses/lisa2000/sld066.htm (2016-04-30)
    Open archived version from archive


  • Analysis: Limiting User Login
    check their mail Obviously this is just too often But what is a more reasonable number We believe that it is reasonable to insist that users wait at least five minutes between login attempts and since we donít really want

    Original URL path: http://www.shub-internet.org/brad/papers/dihses/lisa2000/sld073.htm (2016-04-30)
    Open archived version from archive

  • Analysis: Mailbox Directory
    POP3 protocol requires mailbox locking if properly implemented this shouldnít pose that big of a problem Switching to message locking seems excessive and unnecessary for POP3 environments especially since it creates far more synchronous meta data operations in turn for

    Original URL path: http://www.shub-internet.org/brad/papers/dihses/lisa2000/sld078.htm (2016-04-30)
    Open archived version from archive

  • Implications
    IMAP you need message locking anyway so the use of a mailbox directory format is natural although I still believe that file locking with a mailbox directory requires an excessive amount of work and should instead be pulled into a

    Original URL path: http://www.shub-internet.org/brad/papers/dihses/lisa2000/sld081.htm (2016-04-30)
    Open archived version from archive

  • Scaling: Future Improvements
    It reduces things that we donít care so much about disk storage requirements and it increases the things we do care most about synchronous meta data I O The point of bringing this up now is to show you that

    Original URL path: http://www.shub-internet.org/brad/papers/dihses/lisa2000/sld087.htm (2016-04-30)
    Open archived version from archive

  • Scaling: Future Improvements
    them Of them how many have large databases on them Now sum the overall reliability across those relatively few machines with large databases Now lets go back to the total number of hosts on the Internet ó how many do you think have filesystems on them Pretty much all of them right Now sum the overall reliability of those filesystems across all those machines Now letís compare these two numbers

    Original URL path: http://www.shub-internet.org/brad/papers/dihses/lisa2000/sld089.htm (2016-04-30)
    Open archived version from archive

  • Scaling: Future Improvements
    blown RDBMS provides you can pick and choose the parts that work from both camps and you should still be able to put something together that will work reasonably well After all there are plenty of INN and Diablo news servers out there that are handling a full 200GB full feed these days with similar technologies and they seem to work pretty well So long as you build in enough

    Original URL path: http://www.shub-internet.org/brad/papers/dihses/lisa2000/sld091.htm (2016-04-30)
    Open archived version from archive

  • Scaling: Future Improvements
    start doing integrity checks on every read operation and get into expensive ìvotingî schemes etcÖ My belief is that this is not an integrity issue but more like doing RAID 1 mirroring at the filesystem level and that you can

    Original URL path: http://www.shub-internet.org/brad/papers/dihses/lisa2000/sld092.htm (2016-04-30)
    Open archived version from archive



  •