presentations

Presentations
Log | Files | Refs

2011-FSW.slides.html (4667B)


      1 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" 
      2 	"http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
      3 
      4 <html xmlns="http://www.w3.org/1999/xhtml">
      5 <head>
      6 <title>Scalability &amp; Paranoia in a Decentralized Social Network</title>
      7 <!-- metadata -->
      8 <meta name="generator" content="S5" />
      9 <meta name="version" content="S5 1.1" />
     10 <meta name="presdate" content="20110600" />
     11 <meta name="author" content="Carlo v. Loesch" />
     12 <meta name="company" content="symlynX.com" />
     13 <!-- configuration parameters -->
     14 <meta name="defaultView" content="slideshow" />
     15 <meta name="controlVis" content="hidden" />
     16 <!-- style sheet links -->
     17 <link rel="stylesheet" href="http://www.psyc.eu/papers/ui/psyced/slides.css" type="text/css" media="projection" id="slideProj" />
     18 <link rel="stylesheet" href="http://www.psyc.eu/papers/ui/psyced/outline.css" type="text/css" media="screen" id="outlineStyle" />
     19 <link rel="stylesheet" href="http://www.psyc.eu/papers/ui/psyced/print.css" type="text/css" media="print" id="slidePrint" />
     20 <link rel="stylesheet" href="http://www.psyc.eu/papers/ui/psyced/opera.css" type="text/css" media="projection" id="operaFix" />
     21 <!-- S5 JS -->
     22 <script src="http://www.psyc.eu/papers/ui/psyced/slides.js" type="text/javascript"></script>
     23 </head>
     24 <body>
     25 
     26 <div class="layout">
     27 <div id="controls"><!-- DO NOT EDIT --></div>
     28 <div id="currentSlide"><!-- DO NOT EDIT --></div>
     29 <div id="header"></div>
     30 <div id="footer">
     31 &nbsp; Scalability &amp; Paranoia in a Decentralized Social Network
     32 </div>
     33 
     34 </div>
     35 
     36 
     37 <div class="presentation">
     38 
     39 <div class="slide">
     40 <h2>Scalability &amp; Paranoia in a Decentralized Social Network</h2>
     41 <ul>
     42 <li>Carlo v. Loesch (symlynX)<br/>
     43 <li>Gabor Toth (PSYC)<br/>
     44 <li>Mathias Baumann (PSYC)<br/>
     45 </ul>
     46 </div>
     47 
     48 
     49 <div class="slide">
     50 <h1>Pitfall: Start With Something</h1>
     51 <h2>Cross That Bridge As We Get There?</h2>
     52 <ul class="incremental">
     53 <li>Let's just get started with something!</li>
     54 <li>Pitfall! Historic Examples:</li>
     55 <li>HTTP! HTTP/NG?, SPDY!?</li>
     56 <li>IRC! ...? XMPP!?</li>
     57 <li>XMPP! ...? JSON!?</li>
     58 </ul>
     59 </div>
     60 
     61 <div class="slide">
     62 <h1>One Too Many</h1>
     63 <h2>Multicasting for Scalability</h2>
     64 <ul class="incremental">
     65 <li>social interactions are one-to-many or many-to-many</li>
     66 <li>HTTP is one-to-one</li>
     67 <li>round robin distribution is slow (SMTP)</li>
     68 <li>IP Multicast doesn't do the job (router table overflow)</li>
     69 <li>IRC and NNTP do/did multicast, but had other problems</li>
     70 </ul>
     71 </div>
     72 
     73 <div class="slide">
     74 <h1>One Too Many</h1>
     75 <h2>Multicasting with XMPP?</h2>
     76 <ul class="incremental">
     77 <li>70% of S2S XMPP messages is presence updates (5 years ago)</li>
     78 <li>XMPP has limited support for one-to-many communications</li>
     79 <li>XMPP can be improved, but: trust problem with multicast</li>
     80 </ul>
     81 </div>
     82 
     83 <div class="slide">
     84 <h1>One Too Many</h1>
     85 <h2>Multicasting with HTTP?</h2>
     86 <ul class="incremental">
     87 <li>fundamentally feasible</li>
     88 <li>unnatural: HTTP is not bidirectional</li>
     89 <li>requires trust in a federated architecture</li>
     90 </ul>
     91 </div>
     92 
     93 <div class="slide">
     94 <h1>Don't Trust Virtual Machines</h1>
     95 <h2>Commodity Servers are VMs</h2>
     96 <ul class="incremental">
     97 <li>vulnerable cryptography</li>
     98 <li>controlling system accessible by observers</li>
     99 <li>memory can be monitored</li>
    100 <li>automated monitoring of the FSW</li>
    101 <li>anti-terror legislation possible</li>
    102 </ul>
    103 </div>
    104 
    105 <div class="slide">
    106 <h1>Privacy vs. Paranoia</h1>
    107 <h2>How Much Privacy Is Enough?</h2>
    108 <ul class="incremental">
    109 <li>just to the intended recipients (e2e encryption)</li>
    110 <li>packet size padding (unobservability)</li>
    111 <li>forward secrecy</li>
    112 <li>private subscription lists (not on a server)</li>
    113 <li>robust and resilient against attacks</li>
    114 </ul>
    115 </div>
    116 
    117 <div class="slide">
    118 <h1>Our Conclusion</h1>
    119 <h2>For the web to become seriously social, it needs a native, open and free
    120 backbone technology that enables communication beyond the one-to-one scenario.</h2>
    121 </div>
    122 
    123 <div class="slide">
    124 <h1>Possible Solution Part 1</h1>
    125 <h2>New Architecture?</h2>
    126 <ul class="incremental">
    127 <li>"Enhanced" P2P with servers as agnostic routers</li>
    128 <li>Options: GNUnet, Maidsafe, A3, Tonika</li>
    129 <li>Less possibly: I2P, Diaspora</li>
    130 <li>Features: Unobservability, End-To-End ...</li>
    131 <li>What's missing?</li>
    132 </ul>
    133 </div>
    134 
    135 <div class="slide">
    136 <h1>Possible Solution Part 2</h1>
    137 <h2>Things to add on top</h2>
    138 <ul class="incremental">
    139 <li>multicast routing option</li>
    140 <li>social trust metrics</li>
    141 <li>... applied for routing</li>
    142 <li>flexible payload syntax</li>
    143 <li>standard formats may be portable</li>
    144 <li>web interface still likely</li>
    145 </ul>
    146 </div>
    147 
    148 <!-- div class="slide">
    149 <h3>Thank you.<br/>Questions? Other points of view?</h3>
    150 </div -->
    151 
    152 </html>