Our client Zultrax runs both the Gnutella network as its own Zepp network.

On Gnutella we included a feature in our recent versions we are not so satisified with but it increases our speed significant.

Like any other Gnutella client Zultrax mentiones its client name in the header which it sends when a connection is made. We do not do that anymore. Instead we mention randomly either our own name, the name (and a imitation of its header) of another client randomly chosen or the name (and an imitation of its header) of the client where the contact is made with. This means that in about 50% of the cases Zultrax hides its identity and pretends to be another client.

Strangly enough this results in more succesfull attempts to make connections, more searchresults and faster downloads!!!!!

As far as we can see this can only mean that there must be a major re Gnutella client which give priority to other nodes on the condition that they use that same client.

In case you are also developing on a Gnutella client you might use this knowledge and pass it by the same way we did.

Did anyone else notice this phenomea?

You mean the vendor ID spoofing? Yeah and it sucks. If Zultrax does it, it sucks too. Get some ethics people.

__________________
Liens d'intérêt /Links of interest:
Gnutellaforums en français /The House's rules you have to respect / First search the forum, then create a thread / Free software alternatives! - Logiciels alternatifs gratuits!/

Zultrax does nothing that sucks.

What sucks is the use of a vendor ID or a connection header to benefit your own client above other clients. It is theft of bandwidth. If that becomes a habit it would be rather destructive for the network.

Each client can implent its own mechanism deciding how and how fast to reply to requests of others but it should not matter which client the other is using. We will not allow that our users are being put behind while they are using a totally fair client themselves.

When we would notice this abuse stops we would also remove our defence against it.

You obviously never followed the gnutella saga over the years... Vendor preferencing permitted gnutella to survive by dispreferencing greedy clients which compose the majority of clients.

__________________
Liens d'intérêt /Links of interest:
Gnutellaforums en français /The House's rules you have to respect / First search the forum, then create a thread / Free software alternatives! - Logiciels alternatifs gratuits!/

It is apperently also used for other purposes because Zultrax is in no way Zultrax is a greedy client but it still would get dispreferenced.

Futhermore it is not a very efficient way to reach that goal because it can easily be avoided as I alredy described.

An alternative and more honest way to shut up greedy clients might be actually looking at the traffic a connection is generating. Compare the number of queries, the number of queries with hopstaken=1 and the number of query responses with a recent average and you easily filter out the ones that are greedy. We also include the number of pings in that because some send truckloads of them around with no clear reason.

Zultrax is right, we need a much more secure way for vendor preferencing. Maybe some algorithm using cryptographically strong encryption like BearShare does, so other clients cannot spoof other vendor IDs.

Fortunately, there are not enough Zultrax clients to cause any damage in spoofing vendor IDs but I believe we shouldn't give clients advertising functionality that they just do not have any chance at all.