Project

General

Profile

Protocol core-core Version 3 » History » Version 51

Greg Burri, 08/02/2009 10:26 AM

1 47 Greg Burri
h1. Protocol core-core
2 1 Greg Burri
3 46 Greg Burri
This page describes the protocol used by Aybabtu for network communication between cores.
4 1 Greg Burri
5 49 Greg Burri
See source:application/Protos/core_protocol.proto for message details.
6 1 Greg Burri
7 21 Greg Burri
!network_protocol.png!
8 43 Greg Burri
9 51 Greg Burri
h2. Downloading thread
10 29 Greg Burri
11 39 Greg Burri
There is _NbDownloadThread_ threads. Each thread will choose a free chunk and download it. A peer cannot have more than one connection per other peer. A peer which a download is active is set as not free.
12 23 Greg Burri
13
<pre>
14 27 Greg Burri
F : the list of all unfinished download
15
C : A list of chunk candidates (empty at start)
16
17 33 Greg Burri
- Take the first file in F -> f.
18
Loop :
19
   If the hash chunks of f are unknown :
20 37 Greg Burri
      If there is already a peer asking for hashes :
21
         - Take the next file in F -> f.
22
         - continue
23
      Else
24
         - Ask the hashes to the source (see the message 'GetHashes').
25 41 Greg Burri
   For each chunk not downloaded and not being downloaded of f -> c :
26 27 Greg Burri
      If c has a free peer :
27 37 Greg Burri
         - Add c to C
28 27 Greg Burri
   If C isn't empty :
29 30 Greg Burri
      - Take the best chunk in C based on peer.downloadRate.
30 1 Greg Burri
        If there is more than one best chunk then choose one randomly.
31
      - Grab the lock for the peer
32 42 Greg Burri
      - Request the download (see the message 'GetChunk')
33 33 Greg Burri
      - Take the first file in F -> f.
34 34 Greg Burri
   Else :
35 33 Greg Burri
      - Take the next file in F -> f.
36 23 Greg Burri
</pre>
37
38 51 Greg Burri
h2. Updating peer to hash association
39 23 Greg Burri
40 50 Greg Burri
This loop will periodically grab the _NumberOfChunkToUpdate_ first chunks and request to all other peers what chunk they have.
41 1 Greg Burri
42 40 Greg Burri
If there is no chunk available the message will be send without hashes. This message is also used to know the peer on the network.
43 37 Greg Burri
44
<pre>
45
F : the list of all unfinished download
46
C : list of latest chunk hashes sent
47
48
Loop :
49
   Time.now -> t
50
   Empty C
51
   n : number of chunk = 0
52
   While take the next file in F -> f
53
      If C.count >= 'NumberOfChunkToUpdate' :
54
         - Break
55
      If we know the hashes for f :
56
         For all chunks of f -> c :
57
            - Add c to C
58
      Else :
59
         - Break
60
   - Send the list of the chunk to everyone. (The received responses will be compared against C)
61 1 Greg Burri
   - Wait 'TimeBetweenChunksUpdate' - (Time.now - t)
62 37 Greg Burri
         
63
</pre>
64 23 Greg Burri
65 46 Greg Burri
h2. Parameters
66 8 Greg Burri
67 10 Greg Burri
Most parameters will be stored as constants in their component.
68
69 9 Greg Burri
|_.Name|_.Value |_.Unit|_.Explanation/Comments|
70 19 Greg Burri
| _IMAliveFrequency_ | 0.1 | [Hz] | The message _IMAlive_ is sent each 10s. |
71 14 Greg Burri
| _ChunkSize_ | 33554432 | [Byte] | (32 MB) |
72
| _LanSpeed_ | 52428800 | [Byte/s] | (50 MB/s) |
73 17 Greg Burri
| _TimeRecheckChunkFactor_ | 4 | - | If the time to download a chunk exceeds _TimeRecheckChunkFactor_ * _ChunkSize_ / _LanSpeed_ the peers are rescanned to choose a better one (better download rate). For example : 4 * 32 / 50 = 2.6 second |
74
| _SwitchToAnotherPeerFactor_ | 1.5 | - | During the recheck of the peers to choose a better one the other download rate must be above the current * 1.5. |
75
| _DownloadRateValidTime_ | 1500 / _LanSpeed_ = 300 s | [s] | (5 min). The download rate of a peer is only valid during this time after its last update. After that its rate is set to the maximum. | 
76 15 Greg Burri
| _TimeBetweenChunksUpdate_ | 500 / _LanSpeed_ = 10 s | [s] | Each 10 s a daemon will scan the _NumberOfChunkToUpdate_ first chunks from the first queued downloads to know which peer has which chunk. |
77 20 Greg Burri
| _NumberOfChunkToUpdate_ | 100 * _LanSpeed_ / _ChunkSize_ = 156 | - | The first 156 hash chunks (3 kB) will by send periodically to each peer. See message _HaveChunks_ in the proto file. |
78 19 Greg Burri
| _FindTimeout_ | 15 | [s] | A search request is abandoned after 15 seconds from the start of the request. See message _Find_. |
79 25 Greg Burri
| _GetHashesTimeout_ | 200 | [s] | Timeout for the message _GetHashes_. |
80 19 Greg Burri
| _HaveChunksTimeout_ | _TimeBetweenChunksUpdate_ = 10 | [s] | Timeout for the message _HaveChunks_. |
81 1 Greg Burri
| _GetEntriesTimeout_ | 10 | [s] | Timeout for the message _GetEntries_. |
82
| _NbDownloadThread_ | 3 | - | Number of concurrent downloading thread. |
83 45 Greg Burri
| _BufferSize_ | 1048576 | [Byte] | (1 MB). This buffer is used when reading or writing a file. |
84
| _TCPSocketBufferSize_ | 33554432 | [Byte] | (32 Mo). Used when data are received. See _QAbstractSocket::setReadBufferSize_. |
85
| _TCPPeerConnectionKeepAlive_ | 600 | [s] | (10 min). There's can be one or more connection to a certain peer, these connections are kept alive for a period of time. For each known peer there is a pool of connection.  |
86 8 Greg Burri
87 7 Greg Burri
h2. Core <-> GUI