‘OpenSim-in-a-box’ pt2: enabling Hypergrid

OpenSimulator introduces a very cool concept called Hypergrid which is explained in more detail here.  In general it means that you can link regions from different grids together. The benefit is that you can teleport to regions from completely different grids, wearing your assets and inventory.

To make it easy for OpenSim-in-a-box users it is hypergrid-enabled by default, so since it is turned on at start up you only need to link the regions together:

• start up the region
• start up your viewer and log into your grid
• ‘set home to here’ in the ‘world’ (this is important! or you can’t come back easily)
• now on the OpenSim console enter
  link-region 1002 1002 ucigrid04.nacs.uci.edu 9003
  to set the UCIgrid region to position 1002, 1002
• if you open the map now, you see a new region north-east
• teleport there by clicking on it (again: make sure you set your home position first)
• now you are on the UCI Hypergrid!
• to come back you need to teleport home

The regions ‘see each other’ only on the map, not inworld.  So you can’t simply fly over there, you need to teleport.

• To unlink the region you enter
  unlink-region <regionname>
  on the OpenSim console (in this particular example unlink-region “Gateway 3000” )

Of course you can also start up two (or more) OpenSim-in-a-boxes then you always have the chance to link back so that all regions are linked to each other. The coordinates should be consistent, though.

Regarding available Hypergrids: You can find a list of HyperGrid Nodes here but I am not sure which ones are active. We checked ucigrid04.nacs.uci.edu:9003 and it was working.

‘OpenSim-in-a-box’ pt1: setting it up in the cloud

So, how do you install the OpenSim-in-a-box into the cloud?

• goto the AWS console
• open the ‘Launch Instance’ dialog, select the public images tab ‘Community AMIs’
• search for OpenSim-in-a-box, and select the latest version

• run 1 small instance with your keypair and the default security group

• optional: get elastic IP
• optional: get a dyndns (or similar) entry that point so the elastic IP

Then you switch to your favourite ssh client and

• log in via ssh / putty
• wait for ‘screen’ (this might take a minute)
• start ./bin/start_opensim.sh
• login via Hippo or Second Life viewer

Important: open the right ports for freeswitch and OpenSim

Yes, these are quite few (mostly freeswitch) but you need to do this only the first time. You do this in the security group section of the AWS console.

tcp: 22 for SSH
tcp and udp: 9000-9005
[Provisioning for 6 regions on 9000 – 9005]#

TCP

22 (for ssh)

UDP
16384 – 32768

TCP and UDP
5060
5070
5080
64738,64739

9000

(or 9000 to 9000+N with N= number of regions, provided you entered these ports while creating the regions)

[EDITED missing ports]

‘OpenSim-in-a-box’ pt 0: voices in the Amazon cloud

“Voice enabled Virtual Worlds for 10 cents an hour, ladies and gentlemen!”

For all of you who know how to handle Amazons EC2 – you only need this:

ami-cb7a9ca2

go and get it!

For everyone else I’ll explain what this is all about. In short: a ready-to-run voice enabled OpenSim without (too much) configuration hassle.

As I announced previously we today publish the ‘OpenSim-in-a-box’. The first version of this project is the idea to have a virtual world launch ready including voice support. You can use it to simply try out the OpenSimulator experience or to have a virtual meeting, conference, even concerts for more experienced users. After you are done you simply pull the plug on the instance and it is gone. Next time you need you start with a fresh instance and everything is exactly as it was from the start. It is not connected to a grid and non-persistent on purpose. However, it can be connected to a HyperGrid.

In this version OpenSimulator including voice chat without having to worry about source code, versioning or compilation.

It does however require a couple of things:

• you will need an Amazon Web Services account (which is simple to do if you ever bought something – a book comes to mind – at Amazon)
• you should be able to handle basic OpenSimulator server commands
• you should know how to handle either the Second Life Viewer 1.21 or the Hippo Viewer 0.5 or higher.

Note that you don’t need an SL account – probably everyone who can handle the viewer will have one, but you don’t need the account credentials for OpenSimulator.

It whole idea is based on the fact that setting up an OpenSimulator (a virtual world application server) with Freeswitch (a telephony plattform) can be somewhat tricky, although it’s documented very well (there is a german version, too). Especially thanks to both communities and developers writing the Freeswitch bridge (Rob Smart et al.)  we have a working albeit simple voice chat solution.

Some things you should know:

• for some reason the Linux ‘screen’ takes it’s time to start up
• in the OpenSim Region start-up sequence you need to enter the numerical IP adress of the Amazon instance as the external IP address (or even better: get an elastic IP and a dynamic DNS service like DynDNS)
• Amazon will charge you 10 cents per hour (in the U.S., and 11 in Europe)
• voice support here is very different from voice support in Second Life; everyone in the region hears you can vice versa. The different avatars are practically in one telephone conference.

For future versions we plan to have some more features (not necessarily in that order):

• some initial content for the region
• most hopefully the possibility to be able to phone into the OpenSim (yes, from a real phone)
• web interface

In future postings I’ll explain how to set up Amazon in detail, how hypergrid works and how to have external chat enabled.

So, please try ‘OpenSim-in-a-box’ and give me some feedback on how it works for you! If there are problems, I’d be glad to help.

Dear M Linden …

or: a concept for Linden Lab.

… here’s an open letter.

I talked to several people about the following and my feeling is that this could be of some interest so I figured I better write it down.

I don’t want to sound like a used car salesman here but if I could make a suggestion to M Linden (you are actually quite responsive so I guess I can)  it would be the following:

Divide the existing Linden Lab construction into several companies.

To explain why I first enumerate these companies with some abbreviations and names:
1) LL: The ‘original’ Linden Lab (with focus on ‘Lab’) responsible for supplying technologies for persistent 3D virtual worlds.

2) SLGF: The Second Life Grid Foundation. The world that started it is the biggest grid, but should not be the only one.  These are the guys who run the Second Life Grid. They maintain a free, persistent virtual world without age limitations and free citizenship. I think this is close to the vision Phil Rosedale had initially.

3) LLPS: Linden Lab Professional Services, the guys who consult all the grids out there that are technically based on LL grid server. One of these grids is the Second Life Grid, but there could be/should be be several others.

Like the …

4) AGC: Adult Grid Corporation, the guys who run the adult grid. This would not be connected to Second Life (except that they use the same technology). Probably not too good of a name, I know, somebody can do better. I’d like to stress the following here: I think it’s important not to white list the innocent (that’s why I think the teen grid is not a good idea) but rather separate the adult content.

Why would this make things different?

1. The areas that these companies work in need to be addressed by completely different business models. You need to do different things when you run a service compared to provide the technology let alone make money from communities.
2. Service providers doing Internet business could walk up to their customers and sell Linden Lab technology without even mentioning Second Life – which would be badly needed in Germany for example.
3. The importance of different clients – preferably white labeled would be obvious then. And Phil mustn’t do this with his own league of extraordinary developers but as part of LL work (see also my note here).
4. The goals of Linden Labs (i.e. the different units) would be transparent and believable. Nobody needs to wear many hats.

Let me make some examples:

• IBM would talk to (the new, independent) Linden Lab to buy the technology for their customers since they usually would do consulting on technology on their own since they want to adapt that to there machines (like in hardware) or in their software (like Sametime 3D).
• An effort like OGPX (former MMOX as Infinity pointed out to me) would completely be led by the new Linden Lab which would be an independent  technology provider not a service provider with it’s own interests.
• There would be a distinction between the grid and the technology behind it (which would be very important for the success of the AWG).
• (the new) Linden Lab could think about OpenSourcing it’s server technology easier since there is no dependency into other business models anymore.
• If any of the grids goes down for some reason, the technology provider would still be there. And we wouldn’t have to wait another 7 or 14 years for virtual worlds to arrive.

The thing missing from a technical and a business stand point (apart from the no-brainers like stability) most badly IMHO is a hyper gridding mechanism. I say ‘a‘ hyper gridding mechanism because the one currently in discussion and partly implemented by the great Diva Canto/ Christa Lopes in the OpenSimulator community is not the only possible implementation. And maybe not the implementation that the current Linden Lab wants.

But now Linden Lab doesn’t push Hypergridding enough because it’s not in their business model right now … but you sort of need it for the Immersive Workspace Initiative.

So I guess, Mark, you are wearing different hats in this game now.

Upcoming Amazon EC2 Image featuring Freeswitch VoIP

This is just a quick note that there will be a voice-over-IP enabled OpenSimulator available as a public image on EC2 within the next days.

After some (well, a lot!) of tweaking we are finishing this up right now and it works very well already. The notion is to have a (nearly) out-of-the box solution for virtual conferences with extranet or intranet like character. It tries to save you from the hassle of finding sources, libraries and configurations.

You simply start the image into the cloud and associate an elastic IP to it. Then after you go through the standard OpenSim setup, the voice support is there – you only need to enter the IP address or domain name once in the setup. Then some scripts configure of Freeswitch and OpenSim correctly according to the Amazon Elastic IP provided.

As I said it is as good as ready so if anybody wants to give it a try early then leave a note.