How to improve PirateBox ?

PirateBox is an interesting concept, but it had a lot of limitations based on the fact that the distance covered by a wifi connection is really limited. The mobile devices, able to cover a larger distance also have limitation due to the time needed to transfer any content.

In fact, you can’t imagine to grab content from someone you cross on the street because you might be out of signal before finished to transfer your file. More over due to the actual memory size of the portable devices you can’t imagine to share a lot of stuff on it.

So the idea is great, and the future could change all what I just said. But, today, the reality is that this system is not really usable as a real anonymous and unlimited sharing system.

To improve the system, I would imagine a network of PirateBox, this idea requires to have a larger number of devices but it would allow to share a large amount of data even on restricted memory cards. The idea is that each PirateBox would have a second wifi adapter to connect to another PirateBox. To get the list of its files and share it across this point to point connection. As each of the Piratebox is connecting to a second one we could imagine to build, dynamically a large piratebox network.

The file requests and data transfers would goes from a point to another point without keeping trace of these transfer out of the point to point exchange.

I assume there are some interesting research around this idea as the system has to build a dynamic network, avoiding cycles and optimizing the communications to make the network larger as possible, using a non centralized system to manage all of this.

Anyone to start developing a such stuff ?

PirateBox creation based on TP-Link MR3020

Some days ago I bought a TpLink MR3020 with the objective to create a pirate box and experience this kind of solution. I already tried to do a stuff like this some month ago based on a netgear wifi router having the capability of sharing usb storage. But the system was not easily portable and not extensible.

The proposed solution, based on this low cost router is an interesting opportunity to made the solution mobile.

Continue reading

UTF-8 email, body and title encoding

When sending an email with an application or sendmail, using an UTF-8 encoding, some attributes must be given if you expect the email to be displayed correctly by the reader.

To start, to get the right display in the body of the email, you need to specify the encoding in the header fields by adding :

Content-Type: text/plain; charset="UTF-8" 
Content-Transfer-Encoding: 8bit

The subject is not proceed using this attribute and it must be written in a different way to be interpreted correctly:

Subject: =?utf-8?Q?éssai=20de=20sujet?=

Here, the “?utf-8?Q?” indicates what encoding to be used in the tittle, then it is followed by the title itsleft “Q” indicates that the title is in plain text. The limit of this is that the title can’t contain space, tabs or ? without being escaped firstly. The Title is ended by termination sequence “?=

The other solution to manage title more easily is to use a base64 encoding with the “B” encoding type instead of “Q”

Subject: =?utf-8?B?base64EncodedTitle?=

This way is more easy to encode if your system handle a base64 encoding function.

Hadopi, pourquoi la sécurisation est une fausse piste

Un appel à expérimentation vient d’être publié par  l’Hadopi  concernant les logiciels de sécurisation, arlésienne que le texte prévoit dès l’origine mais qui depuis plus de deux ans ne débouche sur rien. Cet appel est une sorte de demi aveux d’échec d’une idée politique qui comme je l’ai déjà  évoqué ne repose sur aucun fondement technique.

Un premier projet envisageait de mettre le module au sein de la box, comme je l’envisageait dans mon article de 2009, solution que je ne jugeai pas faisable pour des raisons de couts et de standardisation, d’une part, mais aussi de par le fait qu’elle ne prouve aucunement de l’innocence de qui que ce soit dans la majeure partie des cas. Le nouveau projet envisage donc l’utilisation de composant interne au réseau, distribué sous forme d’un logiciel à installé. Voila en gros le cahier des charges de l’Hadopi en manque d’idées.

Voici maintenant en quoi il ne sera jamais possible de prétendre de l’innocence ou de la culpabilité d’un personne avec un tel produits, ni même, ce qui pourrait être acceptable, de la bonne foi du prévenu :

Continue reading

FreeNas and OpenFiler experimentation

 

/!\ Article in progress, not yet finish /!\

As I needed to create a iSCSI share for some VM on a private subnetwork, instead of using a simple NFS server configuration, I was looking to test a NAS distribution. I saw on Internet two different distributions, one based on OpenBSD named FreeNas (here in version 8), the other based on Linux OpenFiler (here in version 2.99). I will test both for creating this share.

Continue reading

Oracle VM – experimentation

/!\ Article in progress, not yet finish /!\

I was looking  to test Oracle Linux since Open World, mostly because I’m curious about OVM and the nice integration with Enterprise manager 12c as a global system to manage hardware, operating system, hypervisor, virtual machine on demand and middle-ware + software.

Continue reading

Activate Masquerading (NAT) on Linux router

To activate NAT on a Linux Box used as a router, just use the following line :

iptables -t nat -A POSTROUTING -o eth0 -j MASQUERADE

eth0 is the network interface able to access Internet directly

Then you can list the NAT entry in iptables with the following command

# iptables -t nat -L

You can get more details with:

# iptables -t nat -L -v

The conntrack tool also help to see what happen in the NAT

# conntrack -L --src-nat / --dst-nat

Latency impact on NFS link

Interesting document about NFS and iSCSI performance over latency, even if it is new a new document, the study made is really complete and interesting. As I was mostly interested on the performance of NFS over a WAN access with a high latency, I would summarize it by concluding that the maximum performance of file transfer over NFS is not so far the following list:

Time to read 128MB / latency (read is the worst case)

  • 10 ms latency requires 200s, max bandwidth is about 640KB/s – 5.12Mbits/s
  • 20 ms latency requires 300s, max bandwidth is about 427KB/s – 3.41Mbits/s
  • 30 ms latency requires 500s, max bandwidth is about 256KB/s – 2.05Mbits/s
  • 50 ms latency requires 800s, max bandwisth is about 160KB/s – 1.28Mbits/s
  • 90 ms latency requires 1600s, max bandwidth is about 80KB/s – 640Kbits/s

For details and much more information, take a look to the source document : http://lass.cs.umass.edu/papers/pdf/FAST04.pdf