[sldev] Opening the server source?

[Adam Frisby]

This is going off on a bit of a tangent from the original discussion, 
but nonetheless is probably rather important background. I suspect LL 
may look in on this but not actually comment any further.
--

This is really up to Linden Lab. It's a case of quantity vs quality of 
customers. LL's going to sell a lot more "regions" if they charge a 
connection fee of $5/month and hosters can come in.

As a very quick example, I was able to get over a thousand regions 
running on a single server using an experimental lightweight version of 
OpenSim - obviously that isnt a practical real world example, but it 
does show that the "per sqm" pricing model is fundementally flawed 
compared to the real costs of hosting.

The actual costs of hosting a region are going to come down to things 
like "What does this region cost to support - how many can run per 
server in this configuration, etc" - and most of this is going to end up 
very consumer focused like the web hosting industry is today.

Now, the economic argument is in favour of opening the grid for one very 
fundemental reason - if LL doesnt do it, someone else eventually will. 
LL doesnt have any patents or similar covering the idea of a virtual 
world and their feature set is not unclonable. In 18 months, by the time 
that LL is ready to open the grid to third parties, there will be 
competitors around and active. Smaller competitors also have the 
advantage of smaller loads and knowledge of where things fall apart from 
the get-go.

Most of us know the names already - the two from China, the one from 
Australia, et cetera. All of them are in alpha or beta now, but in a 
year and a half - they will be public - one of them will be bound to 
make the decision to allow third party hosting, and that will bring 
significant investment into them as large companies are able to control 
their regions to a degree not possible with LL hosting as it is today. 
See MTV or Wells Fargo for that demand. Never underestimate the power of 
commercial investment to spur development either - see the early life of 
Apache for a great example.

Yes, this will be massively disruptive to the land industry - I know 
quite well (being the second largest 'landholder' inworld afaik), but 
I'm also a realist - even despite the disruption, the end result will be 
that Second Life grows far, far in excess of it's bounds today, and if 
it doesnt happen, someone else will do it, and will benefit from the growth.

LL is moving towards being the ICANN of the 3D Virtual World - likely 
what they will sell will be region handles on the main grid, and the 
ability for your users to connect into it from other gridded servers. LL 
will probably also sell things in addition, like asset storage space, etc.

Obviously LL will need to rethink what they host in terms of a cost 
argument (It may end up being that you pick a 'account host' as well as 
a 'land host' where the account host stores your assets for a fee, are 
advertising supported, or supported through their own hosting service)

By shifting the load onto third party servers and distributing things, a 
lot of these load arguments also vanish due to economic imperitives - 
want to provide a reliable service or lower bandwidth bills? setup your 
own asset caches closer to your servers, etc.

LL's real chokepoints that I can see in the long term will be things like:
* Unlimited free inventory storage - this just isnt practical (and has 
been showing signs of this for the last two years)
* Unlimited free asset storage
* etc

LL's benefits are:
* Massive benefits for consulting / services "We invented the darn thing!"
* Classed as a premium host (Know the platform, where scaling is needed, 
can roll out custom features faster & easier)
* Fundementally control the grid and can charge outsiders for a 
connection (thus ensuring a better margin and competitors)
* No need to manage certain scalability elements as third parties manage 
their own.

Lots of things like grid region handle resolution can be scaled in the 
same way the DNS system has been scaled - economic imperitive and highly 
distributed load.

 From a developer perspective there's a lot to be gained as well, 
customised simulators done to specific tasks (eg viewing realtime 
scientific data in collaboration with other people) - the Croquet 
project is a fantastic proof of concept for things that will be 
possible, but the two fundemental questions for opening the grid are -
a) Can Linden Lab Profit long-term from this?
b) Can SL grow, or even survive without third parties long-term?

Taking a 5-10 year look, I dont see SL surviving the "fad" stage without 
it, it may simmer, but it eventually will cool unless it becomes a 
standard -- ActiveWorlds these days has around 20 or so concurrent users 
because there hasnt been much interest in the platform, nor practical uses.

</essay>

Adam

Jason Giglio wrote:

> Mike Monkowski wrote:
> 
>> LL gets a benefit because it gets subscribers without having to host 
>> the whole thing.  Private hosts get the benefit of minimal lag, 
>> privacy for confidential information, and the ability to customize the 
>> behavior of their worlds.
>>
>> It's really a question of what business LL wants to engage in.  I 
>> don't think their goal is to manage more and more server farms, 
>> because their 
> 
> 
> For $6700 per server + $1180 per month per server, I think they are 
> quite happy hosting server farms.
> 
> Any sort of open grid is going to decimate the market for linden land.
> 
> To put it in perspective, if you were LL you'd need about 150-200 new 
> premium accounts for each server you don't host, to make up the difference.
> 
> And those accounts would use the bits of the service they are having the 
> most trouble scaling well.
> 
> Not to mention, sims probably couldn't handle assets directly like they 
> do now, since that would give sim owners super-copybot-powers, the 
> ability to grab perfect copies, including script binaries.  That would 
> put even more load on the centralized resources, since the sims can't be 
> trusted anymore.
> 
> -Jason
> _______________________________________________
> Click here to unsubscribe or manage your list subscription:
> /index.html
>