|
|
|
[
Permlink
| « Hide
]
LucianoFaretra Ibanez added a comment - 28/Oct/08 04:50 PM
LL can limit the agent to 30 for OS
Pasting in my technical thoughts from the mostly-less-than-technical MISC-1776 (apologies if this isn't the right place to do this):
Seems like it should be possible to prevent "abuse" (i.e. people loading their OpenSpace sims to an extent that hurts the performance of the others on the same machine) by appropriate fiddling with the operating system or hypervisor or whatever. Talk to the techies, and put in CKRM policies or whatever to ensure that each sim process gets at least its fair share of the CPU. (If it's about disk access or network bandwidth rather than CPU, there are perhaps similar things that would let you do similar things for those.) If this was done, and someone tried to put more onto an OpenSpace sim than their fair share, they'd get enormous lag (and their neighbors would see little or no impact). And then people would learn just what they can and can't do, rather than having to just sort of guess based on the ambiguous hints and changing prim limits that we have now. This would also help with the fact that OpenSpaces sims are turning out to require more than half of a full sim, rather than the one-quarter that was expected: use OS or hypervisor policies to limit them to one-quarter, and people will then be unable to get more resources than you can provide for what they're paying. If this has been tried or considered and failed, it would be very interesting to hear about it. It would reassure us that this isn't just a hasty decision, or driven by some unstated purpose. While I have been able to do it with an oracle database I have never tried with a host to monitor resources.
What am I talking about? Well I built a process control monitor system for oracle databases that collected the key metrics over time on a stable run and then I used six sigma (+/- 3 std deviation math and some rules really) to set control limits. If any measure fell within the rules for trending out of control or gone out of control then it would page me. Saved much unplanned outages from happening. Even detected what turned out to be a dead watch battery on a raid controller board lol. So if it is possible to take cpu, memory, and network measurements over time for how an open space or regular sim is supposed to run then a monitor can be developed to log sims that have issues. Then someone can have a look and if the sim is being abused in a manner that violates a TOS policy then that region's owner can be dealt with. The key is a clearly written policy that can be enforced backed up by enforcement and a monitor tool set that can make the process of enforcement thorough and cost effective. Now if the control limits are determined to be incorrect after some period of time then they can be recomputed. This technique is used in things like EPOS process control monitor systems that ensure the parts in your anti lock brakes work faithfully because they are made within exact tolerances for the entire thermoplastic molding process. So it is not a new thing and has been in use for decades. It is a matter of applying the principles to code for this application. A social solution to what is a social and not technical problem.
I believe that non-profits are not trying to capitalize on any kind of get rich scheme they are just raising lindens for their causes if there is a problem in how they are doing it then LL needs to address those non-profits, which by the way, there are many good causes and work going on in s that benefit thousands and thousands of people. I do not have a open space for my non-profit organization but those that do need to be allowed to continue on with the job at hand which is helping someone else in need by the work they are doing. If in fact there is abuse in sl open spaces with non-profits, then LL needs to single those ppl out not make every other non-profit suffer, it is like your big brother hits you then goes to mom and says he hit me and I did not do any thing and she punishes both of you the same way and you did nothing at all, now is that fair?
may be it s allready listed as bug in jira (don t know how to look for that), some scripts are more laggy after few hours they run , even if they do the same the whole time,
last was a simple playloop function, starting from 0,01 ms ending with 6ms!! even reset didn t changed anything. & this ms slide happens with many scripts. fixing that would resolve much things.. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||