Fees

If i buy land in an auction for $500.oo do i still have to pay lindens monthly fees?
Is there anyway to own more than 512m2 of land in SL without paying more a month?

Yes

and

No

helpful and not helpful

so what does that mean ?

You have to pay tier fees in addition to any cost of the land itself.......

Cost Sq. Meters Amount of Land
US$195.00/month 65,536 Entire Region
US$125.00/month 32,768 1/2 Region
US$75.00/month 16,384 1/4 Region
US$40.00/month 8,192 1/8 Region
US$25.00/month 4,096 1/16 Region
US$15.00/month 2,048 1/32 Region
US$8.00/month 1,024 1/64 Region
US$5.00/month 512 1/128 Region
US$0/month 0 0/128 Region

512 sqm is included in your monthly cost with the premium acct.

Land Fee's suck!!!! Why can't our basic monthly fee be enough. Why must we pay more if we aquire more... The rich are paying more than their fair share, whereas the poor pay next to nothing.

YOU PEOPLE VOTED FOR GILBERT HUMPHREY, AND YOU KILLED JESUS!!!!

HURRUMPH

Why must we pay more if we aquire more...

I can buy a 12 oz. can of Coke at the store for $0.50. Or, I can buy a three liter (101 oz.) bottle of Coke for $1.69. That is exactly how land works in SL. I have to pay more to get more, but I get a volume discount.

Slight volume discounts aside, there are two reasons you pay land fees.

1) Laze-Fare: Throughout history, money has been used as a method of voting. If people like something, they spend money on it and the providers can afford to keep providing it, if something is just there not serving a benefit to anyone it disappears. This way if someone builds their dream land, and other people love it too, then they will express there love by hanging out and letting the land owners collect dwell, or doing business with the owner. If no one likes it, the owner gets no help in keeping it running, which is fair seeing as it is benefiting no one but them. This way we don't need oppressive rules as to what people are allowed to build in order to keep SL beautiful. The economy takes care of it.

2) Cost: Virtual land costs a lot of money to exist. They run on expensive servers, and used even more expensive bandwidth, and require very talented and expensive technicians to maintain them. The system needs, developers, debuggers, management, liaisons, etc. Linden Lab is not funded by the government, nor does it have an endless money supply. But they do have investors who would like to see a return on their investments. Therefore they can not bear the cost of these resources without the help of those who use them. So they *must* charge land fees. If you want a boring, static world with a finite number of never changing areas, and pre-made objects then you can find that at “someWhereElse.com”, but we all know that place sucks.

The good news is that as technical resources constantly come down in price, land fees will either stay the same with better technology, or get cheaper, or both.

LL Doesn't know what they are doing. They need to be introduced to Clustering so they can maximize storage, bandwidth, etc.... Don't believe for a minute that LL's methods of network design are cost efficient. If each Sim is an actual server, they are wasting money, time, and resources.

Hire me on as the VP of Engineering and I'll scrap the existing setup and build a Super Computer. The Univ. of Kentucky built 128 AMD Athlon machines which has the processing power of 471 GFLOPS. http://aggregate.org/KASY0/

LL is just wasting money and could better design SL for cheaper, thus reducing the need for land tiers to fund sims machine by machine. Just build a super computer and make each sim virtual. Or no sims at all.. Maybe even take that idea and break up the network into regions around the country so massive bandwith to 1 location isn't needed.. Cheaper bandwidth to various regional hubs might be more cost effective. GigabitEthernet at various sites might be more cost effective than OCx circuits.

In any case, I'm sure the current network design is not efficient and burning up all LLs money. As for management and such, they can farm that work out to India for pennies on the dollar. I've found companies in India who contract out workers for about $800/mo for 24/7 work from IT design/development to simple customer support. But hey, I'm just a silly man from a Red State, those Blue state people can't understand my English.

lol

Oh the humanity!

Apart from that -- I'm still trying to see the problem - but I think I found a new sig line.

Siggy.

Glad I could add to your .sig lines... Without the Ole Alby Meister, you'd be SigLess..

Ummm, I haven't ever heard of anyone outsourcing management to another country, unless they had operations in that country. That would be worse than having a Harvard business grad run your company.

I know I'm not the best one to speculate or critique Linden's systems, because they have really good people handling that who know better than me, but I don't think you would say these things if you understood their master plan. I do but I'm not going to spoil the surprise. Let's just say that there is a lot more potential here with their expandable architecture than just a super sim computer. If super computers were so great, we would run web servers on them. They have been great for crunching numbers, and pong, but not much else.


Usually when someone says "bah, all you have to do is," they don't understand the whole problem or they have no clue what they are talking about. Because if such an easy solution was best, it would already be in place.

Except when I say it, because I'm just that much smarter than everyone else.
But I bet if you ask "why don't you just do this?" you will get a huge list of reasons. People aren't giving them millions of dollars to not think things through.

Yeah, thats what I'm saying.. Bah!!!

Why isn't a design setup so that when 1 sim becomes maxed because somebody is throwing a massive party/event that CPU cycles from other machines being shifted to handle the load? If you have 200 machines on a rack all interconnected and the party is on machine/sim "Zuni" which is maxed beyond belief and all 199 are sitting idle, then I'd call that a waste of resources.

If all the servers where sharing CPU/Memory/etc in a clustered enviroment, then any one sim which started to become maxed would pull in resources from other sims which are idle. As such, keeping the load, bandwidth, etc down. And when any 1 server isn't causing the physical limits, then the tiering metric can be reworked to be less of a burden on a player. Thus allowing for players to own more land, more prims, and create more content.

Clustering is more than just numbers crunching and running httpd processes. Its about sharing resources across many machines. What is a super computer? Just a box of processors. You can either buy an offical super computer or you can cluster many smaller machines together at a cheaper price and achieve the same effect.

I truly don't by the arguement that if something was better it would have been done. I've seen many times in the corporate world thousands of better ideas and/or ways of doing things. Yet its never done, manly for political reasons. Somebody doesn't want to lose face that their idea sucks, so they rather keep throwing money and time at a bad idea than admit defeat, switch gears, and go down a new path. How do you explain to investors that the millions spent on this current design is not longer effective and we new more millions to switch plans? Not so easy when your playing with somebody else's money. Pride, Greed, Fear, you name it are reasons why bad mouse traps continue being bad mouse traps, even when the plans for a better mouse trap are within reach.

Here is a nice little blurb that backs up what I'm talking about. Have LL explain why sims lock out players if too many max the server?



Source: http://www.newsfactor.com/story.xhtml?story_id=18722
----
First Gaming Grid

In a real-world example of grid computing, IBM and Butterfly.net announced in May that they would soon release a computing grid for the video game industry. Butterfly.net spent two years building the grid, which distributes games across a network of server farms using IBM e-business infrastructure technology.

Massively multiplayer games (MMGs) historically have been run on mirrored servers that essentially duplicate copies of the MMG universe to balance user loads.

While this technique is designed to reduce latency for all users -- so that each set of servers behaves responsively to user actions -- the mirroring technique limits the number of players who can participate at one time in the same game universe.

When load balances increase, the typical MMG response has been to add more servers, copy the game universe and spill the extra load into that new copy.

Now, however, Butterfly.net's grid technology provides "cross-server sentinels" that supports the interaction of millions of players in one world, with server boundaries invisible to players. According to the company, the extension of grid computing to the gaming world lets game developers support a limitless number of users in their MMGs.

There is very good reasons why solutions like this never take off:
- They are unreliable. Crash one, or put one into a spinlock, and voila, you've done it to the whole network
- Implementation costs would be more than the cost of providing beefy hardware to everyone. (nb: all the sim hardware >sim100 is suitable for heavy use. it's just a few troublesome machines in the 1-100 range that are proving a problem)
- This will be a nightmare to undo when LL decides to decouple te grid.

For technical reasons:
- Pipelining instructions and memory over a network is both dumb and prone for very very very high inefficiency. Consider that L1/L2 cache is in the 30gbps range, while your PCI bus (which your network card is going through) tops out at 640mbps, assuming you have gigabit (which LL doesnt), in which case it tops out at below 100mbit. Even comparing to traditional RAM looks a whole lot better than networking. Additionally: networks have inefficiency which would kill this system. You would need extremely high-end network switching gear to handle this, and that 100mbit is going to be shared between every computer trying to send stuff to every other computer. (so this becomes a 1/n efficiency problem)
- This requires a application which paralellises easily (anything that access memory a lot doesnt classify.), and requires using fairly expensive hardware to achieve good results from. (Itaniums spring to mind)

-Adam

Where were you when I was trying to explain exactly that to Morgaine Dinova, who ignored whatever facts I threw at her, saying she had a phd and knew very well that SL could be easily rearchitected into a dynamic resource allocation system?

LOL Eggy, that just an example of how most of the time ... whoever talks the most and claims to have the most knowledge in fact is just compensating for their own insecurity. Especially if they keep saying they have a PhD, or have to repeatedly drive home the fact they are a <insert some smart sounding tech job here> in RL, or otherwise overly display an air of "intelligence" about them.

It's one of the ways you can seperate the veterans from the upstarts We dont talk about how smart we are. 'Cause experience teaches us that we really aren't.

For SL, the short answer is that it could have been done if it was designed that way from the beginning but it would have been much much more expensive & not feasible for this sort of application. The potental return on investment wouldn't have been there, and ultimately that is the most important part of any system design.

-AP

I could drag up the post if you want? =P

-Adam

I am a sanitation engineer and I have found a way for the SL grid to use tachyons to dynamically balance load prior to the load ocurring. I'd post the explanation here, but sending ASCII via toggle switches from my IMSAI 8080 is slow...

LL has a posting for a Sys Admin.... opportunity to make these changes on the inside.

Oh yeah, I'd love to get that page at 3am. Sign me right on up!

Your making excuses and not acknowledging the facts. 100 Mbps won't cut it locally, you need a GigabitEthernet Network and switches to make it all come together. Is it expensive? Yes. Is it better than the current network design? Yes. LL didn't build for scale. Now its time to switch over to a network design which scales..

PhD? Who needs that? LOL.....

You didnt read my post did you?

Gigabit will top out at 640mbps, this is unavoidable due to the maximum capacity of the PCI bus. Because SL's server is a memory-bottlenecked application. This means it's going to be restricted by how fast it can shuffle things in and out of cache (and to a lesser degree ram). At the moment, 30gbps is about right. 640mbps most certainly will not be.

Edit:

And let's not forget congestion. If we are shuffling between groups of say 50 sims, this means those 50 machines are going to be competing among themselves for that 640mbps, because they all need to access a common memory area. (hence the 1/n problem). This is an extremely difficult task, as evident by the palty efficiency rates of the world's top supercomputers which are engineered from the ground up (hardware wise here) for this kind of problem.

-Adam

Additionally, this network wont scale. As I said, it's a 1/n problem, - your efficiency will get significantly worse as you add more machines to the cluster. This is becuase all the machines are going to be competing among themselves for memory, and other IO access, and let us not forget - these machines have to stream to the world too: this means all their bandwidth cant be clogged sending data back and forward among themselves, which is what you are proposing here.

-Adam