Vista Collaboration

Very exciting stuff.  One concern that pops up though - These pieces of toast he referrs to... The boxes that inform me of an invite to play Halo 2: If I'm playing a different game full-screen at the time, will these pop-ups interfere with my game?  Such a mis-click could be catastrophic to the game I'm currently playing (probably Starcraft).

EDIT:  Having watched further into the video now, I just want to say that it was good to focus on gaming, as that's a huge market.  After hearing some of the great stuff this will enable, I really feel this is a key part of Vista that needs more evangelism; I'd heard nearly zilch about it up till this part and I think it'll be one of its key selling points.  Now, as soon as there's a crack for that whole DRM thing I'll probably be putting Vista on my shopping list

Good one...thanks

Charles your wetting my appetite for a going deep video about Vista.... can't wait.

We’re going to be filming some going deep videos for Windows Vista Collaboration.  What questions do the 9ers have for the team?

Windows Collaboration technologies:

PNRP (Peer Name Resolution Protocol) - "Serverless DNS".
PNM (People Near Me) - Collaboration with people on your local subnet.

Session Invitation - send and receive invites to your contacts.

Mesh, Grouping and Graphing – Long running group collaboration.

Peer Channel - Group replication.

Windows Collaboration Experience - Project your desktop to people’s computers around you. (Start->All Programs->Windows Collaboration)

- Ernie
Window Collaboration Technical Evangelist

The main thing i want to see is come crazy demos.

One question i have is what the latency would be like for games?

Surely having to replicate the packets would increase the latency? 

I'm really interested in the gaming side of this technology. Can you explain more how games would use this technology?

Cheers

Great video! Is this stuff going to be included in Office12? It would be a killer feature to include ad hoc collaboration in OneNote.

Let us in on your FolderShare and Avalanche plan.

I hear MS has a IPv6 to IPv4 translation service, if I didn't misunderstood its function, there's gonna be a lot of traffic going through it. Can the service handle the traffic?

Hm, MS FolderShare website running on PHP, weird...

Will Windows Update make use of PNM, so if someone on my subnet has already downloaded the update I get it over the lan?  (The checksum should still be got from the Microsoft server!)

 

What about a general system for sharing downloads from MSND etc, without needing any admin from the network manager?

Ian Ringorse
ian@ringrose.name
email address on website

re: suggested question.

Would it be possible for future versions of Windows mobile (PDA, cellphone) be able to participate in this type of p2p collaboration (PNRP or PNM)? ie. in the 'desktop broadcast' scenario when showing a powerpoint presentation directly on other PC's, could some participants be viewing the contenct on an iPaq? Would windows mobile be announcing & reading 'presence' status in a compatible way? People could use virtual earth mobile for locating each other when trying to meet up somewhere.

Ernie Booth wrote:

We’re going to be filming some going deep videos for Windows Vista Collaboration.  What questions do the 9ers have for the team?

Windows Collaboration technologies:

PNRP (Peer Name Resolution Protocol) - "Serverless DNS".
PNM (People Near Me) - Collaboration with people on your local subnet.

Session Invitation - send and receive invites to your contacts.

Mesh, Grouping and Graphing – Long running group collaboration.

Peer Channel - Group replication.

Windows Collaboration Experience - Project your desktop to people’s computers around you. (Start->All Programs->Windows Collaboration)

- Ernie
Window Collaboration Technical Evangelist

Okay, (since this silly forum makes me do things the hard way...

how is this different from Zeroconf, (http://www.zeroconf.org/) which has been providing PRNP/PNM/etc services for years now and is an open IETF standard? Is this compatible with existing Zeroconf implementations, or will it require network admins to deal with both systems? If it's not compatible, does it use the same ports? Can both systems coexist, or do they conflict? Will printer and peripheral manufacturers have to include PRNP/PNM stacks AND Zeroconf stacks to make sure all their clients can easily connect? (Keep in mind that pretty much all the major printer manufacturers have been shipping Zeroconf support for over three years now.)

if they're the same thing, great. If they aren't, why reinvent the zeroconf wheel, and make it harder to deal with Vista in a heterogenous environment?

People Near Me uses WS-Discovery over IPv6 Multicast (http://msdn.microsoft.com/library/en-us/dnglobspec/html/ws-discovery1004.pdf). Using Ipv6 allows the use of IPv6 link local addressing which avoids some of the more unplesant autoaddressing problems of Ipv4. So yes, its a published standard that also is supported by other parties.

Tripp
Windows Collaboration.

trippparks wrote:

People Near Me uses WS-Discovery over IPv6 Multicast (http://msdn.microsoft.com/library/en-us/dnglobspec/html/ws-discovery1004.pdf). Using Ipv6 allows the use of IPv6 link local addressing which avoids some of the more unplesant autoaddressing problems of Ipv4. So yes, its a published standard that also is supported by other parties.

Tripp
Windows Collaboration.

And what do you do in situations where people have turned off IPv6 to avoid the timeout delays cause by it when you're in a solely IPv4 environment?

Zeroconf seems to be a replacement for IP on the local network vs.  PNRP is a DNS like service that works across the Internet.

Ernie Booth wrote:

Zeroconf seems to be a replacement for IP on the local network vs.  PNRP is a DNS like service that works across the Internet.

Well, actually, Zeroconf works fine outside of the local network, and it's not a replacement for anything. If you read the docs, it's just a way to get the kind of services that IPv6 can provide natively out of IPv4. It's nothing more than what it says. Zero configuration.

Again, is there any real information on how well Vista's version of Zeroconf works with the real one? If you have printers that support Zeroconf, will Vista work with that, or is MS ignoring it/NIH'ing it? Will things like PNRP cause problems with Zeroconf services?

I'm really quite familiar with Zeroconf, but trying to find clear, concise info on Vista that isn't either API or Marketing fluff is more difficult than it's worth at the moment. hence my questions about interoperability with existing standards, something that should be quickly answerable by the Vista team.

It sounds like the P2P api is WIN32.  Is there a managed version of this API? and if not, when will we have it??  This technology seems like it was made for .NET!

I stand corrected.  I read through the doc, but the explanation of what Zeroconf actually provides wasn’t clear.   As for Windows Vista compatibility it still supports IPv4 it just additionally supports IPv6 so an implementation of Zeroconf should work fine if it worked on XP.  When you say “how well Vista's version of Zeroconf works with the real one” I think you are referring to People Near Me (PNM) correct?  If so then PNM just uses WS-Discovery which is a standard for communicating it doesn’t have anything to do with printers.  Is there a list of printers that support Zeroconf if we have one kicking around I will test out Zeroconf on Vista.

There are not .Net versions of the APIs with Windows Vista, but there are a few community efforts underway.

 

Community .NET wrappers:

http://www.codeproject.com/csharp/PeerNameResolution.asp

http://www.thoughtpost.com/p2p.aspx

Collaboration Product Team Blogs

http://blogs.msdn.com/ravirao/

http://blogs.msdn.com/kevin_ransom/

http://blogs.msdn.com/tparks/default.aspx

http://blogs.msdn.com/noahh/default.aspx

http://blogs.msdn.com/peerchan/default.aspx

 

 

 

This app requires IPv6 to function. Can you detail the problems you are having with IPv6?

Ernie Booth wrote:

I stand corrected.  I read through the doc, but the explanation of what Zeroconf actually provides wasn’t clear.   As for Windows Vista compatibility it still supports IPv4 it just additionally supports IPv6 so an implementation of Zeroconf should work fine if it worked on XP.  When you say “how well Vista's version of Zeroconf works with the real one” I think you are referring to People Near Me (PNM) correct?  If so then PNM just uses WS-Discovery which is a standard for communicating it doesn’t have anything to do with printers.  Is there a list of printers that support Zeroconf if we have one kicking around I will test out Zeroconf on Vista.

Um...if you read all the supporting docs that the Zeroconf page links to, they're quite clear on it.

I wasn't saying that Vista didn't support IPv4, but if the Zeroconf - like abilities of Vista cause problems with Zeroconf itself because, as it seems, no one's thought to test this stuff in a Zeroconf environment, well, that's going to be a problem.

Zeroconf isn't a printer technology, it's just a way to provide zero configuration addressing, naming and service discovery in an IPv4 environment, since pure IPv6 still is a future world. There are a lot of applications and services using Zeroconf. In the printer world, find a reasonably new HP printer and look at the network setup, you should see an mDNS session. Heck, Apple open-sourced their own Bonjour implementation, and even has one for windows, the info is at: http://www.apple.com/macosx/features/bonjour/ Again, none of this is new, or even hidden away all that much.

Yes, zeroconf includes mDNS as part of its protocol suite.  mDNS, like many other protocols (LLNMR, NetBIOS, PNRP, WSD, PNM, SSDP) can perform configurationless name resolution on a subnet.  PNM differs from the others in that it is intended for finding users over the subnet, not just arbitrary names or machine names.  This is a construct more meaningful for many applications than the other protocols mentioned before.  To minimize the number of actual wire-level protocols, PNM was designed to layer on top of an existing subnet protocol, WSD.

What zeroconf does not provide (at least according to any of the docs I have ever seen) is Internet-scale name resolution.  Its capabilities are targeted at the subnet.  PNRP was designed to provide serverless name resolution across the entire Internet.  The fact that the protocol also allows for resolution within the subnet is an added bonus, allowing one protocol to be used by an application regardless of the environment.  Furthermore, PNRP provides a rich security model that is lacking in other deployed name resolution systems.

Regarding MS technologies "Causing problems with Zeroconf", that should not occur.  There is, in fact, no interaction between the cited protocols. 

"One question i have is what the latency would be like for games?"

Depends on whether it is Grouping/Graphing or PeerChannel.  Grouping/Graphing use the replicated database model, which lends itself better to persistent data.  This works for some games, but not for others.  i.e. it is great for state data for turn-based strategy games, but poor for bullet position data for a FPS.

PeerChannel is built on a more socket-like model without the persistence.  This model is better for things like the bullets.

Regarding latency overall, the latency is definately higher than in a server based model (1 hop vs lg(n) hops), but it is not prohibitive in most cases.  Most scenarios involving a limited number of players (say, 32) do not involve very many hops (4 or fewer), so it is not going to be bad.  Of course, it is going to be a lot better if those 32 are all on one LAN, and it is going to be worse if the players are scattered all over the globe.

nhorton wrote:

Yes, zeroconf includes mDNS as part of its protocol suite.  mDNS, like many other protocols (LLNMR, NetBIOS, PNRP, WSD, PNM, SSDP) can perform configurationless name resolution on a subnet.  PNM differs from the others in that it is intended for finding users over the subnet, not just arbitrary names or machine names.  This is a construct more meaningful for many applications than the other protocols mentioned before.  To minimize the number of actual wire-level protocols, PNM was designed to layer on top of an existing subnet protocol, WSD.

What zeroconf does not provide (at least according to any of the docs I have ever seen) is Internet-scale name resolution.  Its capabilities are targeted at the subnet.  PNRP was designed to provide serverless name resolution across the entire Internet.  The fact that the protocol also allows for resolution within the subnet is an added bonus, allowing one protocol to be used by an application regardless of the environment.  Furthermore, PNRP provides a rich security model that is lacking in other deployed name resolution systems.

Regarding MS technologies "Causing problems with Zeroconf", that should not occur.  There is, in fact, no interaction between the cited protocols. 

Well, to be accurate, there are no "people" on a network. There are end nodes and applications, and that is all *any* naming protocol can do. Zeroconf is quite capable of finding "people", just ask anyone using SubEthaEdit or iChat. Works quite well.

With regard to PNM and the rest, will Microsoft be providing ways for non-Windows platforms to implement these services on things like Linux, other Unix, and OS X, the way Apple and others have with Zeroconf and Bonjour?

jcwelch wrote:

With regard to PNM and the rest, will Microsoft be providing ways for non-Windows platforms to implement these services on things like Linux, other Unix, and OS X, the way Apple and others have with Zeroconf and Bonjour?

WS-Discovery is a WS* published protocol, PNM is implemented as any other service is using WS-Discovery.

UPnP is a Device protocol that has been supported since Windows ME, and is supported in XP and Windows Vista and is widely supported in NAT's, Printers, Media servers, and other devices. It is an open standard (http://www.upnp.org)

Zero-config is a combination of several technologies,
    Auto addressing which has been supported since Windows 98,
    mDNS which I believe has been superseded by LLMNR which appears         to be supported in WinCe and Vista

trippparks wrote:

jcwelch wrote: With regard to PNM and the rest, will Microsoft be providing ways for non-Windows platforms to implement these services on things like Linux, other Unix, and OS X, the way Apple and others have with Zeroconf and Bonjour?

WS-Discovery is a WS* published protocol, PNM is implemented as any other service is using WS-Discovery.

UPnP is a Device protocol that has been supported since Windows ME, and is supported in XP and Windows Vista and is widely supported in NAT's, Printers, Media servers, and other devices. It is an open standard (http://www.upnp.org)

Zero-config is a combination of several technologies,
    Auto addressing which has been supported since Windows 98,
    mDNS which I believe has been superseded by LLMNR which appears         to be supported in WinCe and Vista

Does that mean that all you have to do is be IETF compliant, and you'll be a full player in a PNM cloud? Bonjour uses Zeroconf, but there are aspects of it that differ, which is why Apple open-sourced it, so you could see the details of that implementation. If

What about security in Peer To peer?

If I go around giving my IP to people wouldnt that represent a security risk? I mean IP can give alot of information about me and my location, and could expose my machine to Exploits.

How Do you account for this in vista's peer to peer model?


What about Zombie and viral attacks? wouldnt that mean that there could be more Devestating DOS attacks , I mean where does the security go in this case?

How can you balance the two security and haveing peer to peer with all its good stuff?

I'm very interested in seeing how you guys solved the problem of peer networks splitting up into seperate subgroups when key peers disconnect.

We do not advertise your ipaddress outside of your local subnet for people near me. On your local subnet your ip address is fairly well known due to various network broadcasts (ARP, DHCP, SMB,etc).

For PNRP, you register a 2^256 bit number that is only known to people who have either recieved an invite to your group or people on your local subnet.

We use whats known as signature nodes. a node is elected to be a signature node and if a partition occurs the side that doesnt detect the signature will attempt to reconnect (though various means)

It looked amatuerish for a Microsoft initiative.  On the networking side, after long cliched tripelike observations about traditional IPv4 annoyances, which seemed to badly miss the issue of the side the connection is originated from (albiet later addressed indirectly when a cleint P2P program initiates a connection into a mesh,) the issue of resource locators got mauled, and seemingly by a poorly experienced team, though  I did not stick aorund for the entire clip.I dropped out on the issue of mesh names resource location.  As described, first, as a mesh based technology, a routing based names resolution protocol similar to what OSPF is to IP routing would be needed, ESPECIALLY to be highly scalable.  It is specificaly because of the hiegherachical nature of DNS that its recursive queries work so effectively.  A mesh protocol without sohpisticated routing topology and even distributed P2P servers which knowledge of the topology, becomes a cluster (I need to watch my language) of resource location.And on that point, yes, btw, P2P does not mean there are no servers.  It means that EVERYONE is both a server and a client.  The idea of a client app registering into a distrubted non-heiarchical flat naming structure negotiated among low power client machines, often of limmitted bandwidth, has long been Microsoft's downfall it has over and over again had to rectify by moving flat technologies to heigherachical ones for the purpose of scaling.Whether NetBEUI, the network protocal originaly favord for lan segments that did not scale to networks (unroutable/flat,) to browser based SAP network advertising directory services that always failed unless bouyed by a scaled OS/2 Net Bios Name Server or Windows Name Server, even at that it took massive work to evolve that into DDCHP and X.500 directory services such as Active Directory because the peer to peer meshes of the WINS/DNS services of Microsoft did not scale and were flat among themselves, thus PDC's and directory servers.And here comes some developer from Microsoft reinventing "scalable" flat technologies that can support "thousands," forget it, build on scaled hiegherical resource locator servers to find local P2P or pure serversto guide connection/resource requests.  This is they way it has always been found to work over and over again at every layer of the OSI.Rookie mistakes..

Will it work on the server? I know it sounds odd, but I have this scenario where a server would need to send peer-to-peer "proactive" notifications to clients (so the server would be a peer, with a potentially large cloud of clients). The current Windows XP Peer-to-Peer stuff does not work (or at least cannot be installed on) Windows Server 2003, are you planning to allow installation of the P2P infrastructure on the server?

Thanks a lot,

Axel