jacob.eva/Reticulum
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Updated docs

Browse Source
This commit is contained in:
Mark Qvist 2021-05-17 15:57:41 +02:00
parent 0a0b8c1bf8
commit aa93e475a4
2 changed files with 22 additions and 44 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
RNS/Resource.py
View File

@ -31,22 +31,20 @@ class Resource:
SDU = RNS.Packet.MDU
RANDOM_HASH_SIZE = 4
MAX_EFFICIENT_SIZE = 16 * 1024 * 1024
"""
This is an indication of what the
maximum size a resource should be, if
it is to be handled within reasonable
time constraint, even on small systems.
# This is an indication of what the
# maximum size a resource should be, if
# it is to be handled within reasonable
# time constraint, even on small systems.
A small system in this regard is
defined as a Raspberry Pi, which should
be able to compress, encrypt and hash-map
the resource in about 10 seconds.
This constant will be used when determining
how to sequence the sending of large resources.
"""
# A small system in this regard is
# defined as a Raspberry Pi, which should
# be able to compress, encrypt and hash-map
# the resource in about 10 seconds.
# This constant will be used when determining
# how to sequence the sending of large resources.
MAX_EFFICIENT_SIZE = 16 * 1024 * 1024
# The maximum size to auto-compress with
# bz2 before sending.
AUTO_COMPRESS_MAX_SIZE = MAX_EFFICIENT_SIZE

40
docs/source/understanding.rst
View File

@ -449,38 +449,18 @@ automate retransmissions if *Resources* are used.
Resources
---------
TODO: Write
For exchanging small amounts of data over a Reticulum network, the :ref:`Packet<api-packet>` interface
is sufficient, but for exchanging data that would require many packets, an efficient way to coordinate
the transfer is needed.
In traditional networks, large amounts of data is rapidly exchanged with very low latency. Links of
several thousand kilometers will often only have round-trip latency in the tens of milliseconds, and
as such, traditional protocols are often designed to not store any transmitted data at intermediary
hops. If a transmission error occurs, the sending node will simply notice the lack of a packet
acknowledgement, and retransmit the packet all the way, until it hears back from the receiver that it
got the intended data.
This is the purpose of the Reticulum :ref:`Resource<api-resource>`. A *Resource* can automatically
handle the reliable transfer of an arbitrary amount of data over an established :ref:`Link<api-link>`.
Resources can auto-compress data, will handle breaking the data into individual packets, sequencing
the transfer and reassembling the data on the other end.
In bandwidth-limited and high-latency conditions, such behaviour quickly causes congestion on the
network, and communications that span many hops become exceedingly expensive in terms of
bandwidth usage, due to the higher risk of some packets failing.
Reticulum alleviates this in part with it’s *path* discovery methodology, and in part by implementing
*resource* caching at all nodes that can support it. Network operation can be made much more
efficient by caching everything for a period of time, and given the availability of cheap memory and
storage, this is a very welcome tradeoff. A gigabyte of memory can store millions of Reticulum
packets, and since everything is encrypted by default, the storing poses very little privacy risk.
In a Reticulum network, any node that is able to do so, should cache as many packets as it’s
memory will allow for. When a packet is received, a timestamp and a hash of the packet is stored
along with the full packet itself, and it will be kept in storage until the allocated cache storage is
full, whereupon the packet that was last accessed in the cache will be deleted. If a packet is accessed
from the cache, it’s timestamp will be updated to the current time, to ensure that packets that are
used stay in the cache, and packets that are not used are dropped from memory.
Some packet types are stored in separate caching tables, that allow easier lookup for other nodes.
For example, an announce is stored in a way, that allows other nodes to request the public key for a
certain destination, and as such the network as a whole operates as a distributed key ledger.
For more details on how the caching works and is used, see the reference implementation source
code.
:ref:`Resources<api-resource>` are programmatically very simple to use, and only requires a few lines
of codes to reliably transfer any amount of data. They can be used to transfer data stored in memory,
or stream data directly from files.
.. _understanding-referencesystem: