A Fault Tolerant Peer-To-Peer Replication Network

A Fault Tolerant Peer-To-Peer Replication Network

Abstract — We propose a fault tolerant, peer-to-peer replication
network for synchronizing files across multiple hosts. The
proposed topology is constructed by applying existing
technologies and tools to ensure that files are kept synchronized
even after subsequent modifications. One of its main advantages
lies in the fact that there is no central authority to coordinate the
process, hosts are connected in a peer-to-peer fashion, thus
avoiding a single point of failure. Our proposal is intended for
use in networks of personal computers where a small number of
hosts have to be synchronized.

45d6aaf081e8041a8d8a530616f56651?s=128

Radu Potop

July 02, 2010
Tweet

Transcript

  1. A fault-tolerant peer-to-peer replication network Radu Potop Otto Iovanici

  2. Replication network Syncronizes files across computers Data available wherever you

    go Home Work Remote
  3. How do you do it? What we initially wanted to

    do: Star topology network S C C C C client server
  4. How do you do it? But... What if the server

    goes down? S C C C C ? ? ? ? Bye
  5. Peer-to-peer Descentralized Each peer (node) is both client and server

    Mesh topology (fully connected)
  6. Fault tolerant What if a node goes down? The network

    does not suffer
  7. Fault tolerant Nodes can be added and removed Particular to

    our network: permits any file transfer is secure grants ASAP availability of data
  8. Technology SSH secure network public key authentification private / closed

    network needs initial configuration
  9. Technology Unison two-way synchronization Sync Result A B 1 2

    A B 1,2 1,2
  10. Technology Cron Sync every hour Bash script Reads IP list

    Runs Unison Sync IP list between nodes
  11. How does it all work? Node A syncs (in order)

    to B, C and D A B C D 1 2 3 4 Step 1
  12. How does it all work? Each node syncs to every

    other node A B C D 1,2,3,4 1,2 1,2,3 1,2,3,4 Step 2
  13. Characteristics ASAP availability of data Consistent data across nodes Some

    overhead Inneficient for large networks
  14. Demonstration

  15. Thank you