Sim Owners Take Note-- Your Sims Are Sharing Servers

(NOTE: The following interviews were conducted with full knowledge of the people involved. They are reprinted here with their permission. Some slight editing has been performed to place comments in a presentable basis, while fully retaining intent of the reponses).

Question: Did Linden Lab at any time ask your permission to place your sim on the same server as another sim?
April: No not at all
Question: Did Linden Lab ever inform you that your sim had been placed on the same server as another sim?
April: One time a call them about in june or july and the sim crashed and i was told that the sims was being shared and i ask why was that . and that not what i paid for. So they said sim did not like it so they would correct and leave it to run by itself
Question: So they told you your sim would be returned to a single server. Is your sim on a single, dedicated server now?
April: Im really not sure. for what you told me the other day. its being shared with ginko sim
Question: That is correct. One last question: What is your opinion of this matter? Do you feel that Linden Lab has acted properly in these things?
April: I feel like i have been mislead
---------------------------
David: well..I was pissed when i found out [our sim] is sharing on a dual processor server
David: because she paids to have her own server
David: I didn't buy this sim, but it's always been my understanding that each private sim has it's own server and LL has said as much quite a few times
David: I know what [partner] paid for the sim, and what she pays each month, and it's certainly enough to have her own server
David: what happens if both the sims on a single server have alot of traffic one day, both have good events..?
David: They both slow to a crawl
David: If LL thinks a dual processor server can do the work of two single processor servers, they are sorely mistaken
David: and why does LL switch them without asking or informing?
David: Not a good way to treat great customers
David: [my partner] is a bulldog when it comes to the sim's performace..she hunts down active scripts, listeners, etc...
David Valentino: so we definitely know something is up when performace drops drastically
David Valentino: I've been to functuioning events with 65+ people
David Valentino: we hit 20 or so and things get real funky
David Valentino: I don't understsnd..for what private sim owners pay
---------------------------
Question: When you purchased ElvenGlen and Elf Haven, what were you told by Linden Lab regarding server allocation?
Forcythia: I understood that each of my sims would have it's own server.
Question: Did Linden Lab at any time inform you that your sims were to share servers with other sims?
Forcythia: Definitely not!
Question: Did Linden Lab ever tell you that they never place two sims on the same server?
Forcythia Wishbringer: I don't think that question came up, I was satisfied hearing that I would have my own server.
I expected them to live up to their end of that bargain.
---------------------------
[As presenter of this post and to be fair, I answer my own questions]:
Question: When you purchased ElvenGlen and Elf Haven, what were you told by Linden Lab regarding server allocation?
Wayfinder: We were told that private sims were placed on dedicated servers and run much faster than mainland sims.
Question: Did Linden Lab ever request your permission to share servers with another sim?
Wayfinder: No
Question: Did Linden Lab ever inform you that your sims had been placed on a server with another sim?
Wayfinder: No, just the opposite. When we asked about this due to data pointing to that fact, they denied that our sim shared servers with other sims. We were told the only circumstance under which such a thing would happen would be in an emergency when someone else's server shut down, and that that situation would only last until they could set up another server and correct the problem. We were told that was rare.
---------------------------
(end of interviews)

FACTS:
849 sims examined
374 sims were DOUBLE stacked (44.05%)
57 sims were TRIPLE stacked (6.71%)
64 sims QUAD stacked (7.53%)-- This appears to reflect "void" sims. However, on examination, it seems that in some instances two void sims are stacked on two active sims. How much this affects sim performance is unknown.

Total SL sims sharing servers with other sims: 495
---------------------------
THE DISCOVERY

Oct 9/10, 2005, ElvenGlen sim experienced a significant drop in speed-- lag so bad that although there were only 5 avatars on the sim... it was like walking through molasses. When sim speed was checked, Sim FPS had dropped to 10, and Run Tasks ms had jumped from 1.8 to 16-- an extremely high figure.

Sim owners searched for a half hour trying to find the cause, and could find nothing. At that time, we contacted Linden Lab and spoke with a Linden on duty. During the course of this conversation, it was discovered that ElvenGlen had been "assigned to the wrong server" and was sharing a server with two other sims (a transcript exists of this conversation). Because of forum rules prohibiting replication of IM conversations here, we can only paraphrase Linden Lab's side of the conversation. ElvenGlen's side is replicated en total, by permission.

EG: We were of the impression that each sim has its own server. Are you saying that servers are running 2 sims?
(LL replies that some servers are fast enough to do this)
EG: But it's not what we pay for
EG: We pay for one sim, one server
(LL responds that this practice is carried out on most servers, and claims that there is no deterioration of performance because their servers are dual-core).
EG: I know about dual core. We are not paying to share computers between sims.
(LL claims that such is faster than the old region/server ratio)
EG: LL charges $1200 to set up a sim, supposedly on its own server
EG: Dual core is fast (I own one), but the idea is to make sims faster, not to allow LL to double up servers.

========================================

As the owners of ElvenGlen, we were very upset by this information. Why? Because for MONTHS we have been trying to track down causes of "lag", and found no perceivable cause on our sim. We often described the matter to LL as "someone flipping a switch"... causing our sim to lag excessivly. We were told by LL over and over that the problem was NOT server side, that it was our content, our textures, our scripting... although none of these things had changed. We cut content, cut scripting, and the problems still remained. Constant data readings we took correlated NO content/client side cause with lag incidents. Every claim LL made was unsupported by data; in fact all of our data conflicted LL claims.

We spent untold hours (weeks.. months?) trying to track down these lag problems-- only to find that Linden Lab had been withholding one vital piece of information: SIMS SHARE SERVERS.

Following the above conversation, we contacted Linden Lab, and was told that under no circumstance did Linden Lab stack sims "on the same CPU" (interesting wording, since "CPU" is a multi-definition term). Regardless of that wording, one thing must be clearly understood:

*** We were SPECIFICALLY requesting of Linden Lab as to whether our sim shared a server with another sim, and we were specifically told that no, it did not. *** Regardless of word definitions or specific grammatical constructs, that was the issue at hand. We asked Linden Lab if our sim shared a server with another sim, and we were told no, it did not.

In light of the above conversation, we decided to determine the truth of the matter.



THE PROJECT
===========
We called for volunteers from Elf Clan to help us verify server use. These volunteers spent TWO DAYS going to every sim on the grid that we had time to reach, and copying down sim names, server numbers and IP addresses. Although we did not have the time/resources to check every single sim, we were able to examine 849 of them-- a high majority sample.

After compilation and data analysis, the results were astounding:

849 sims examined
374 sims DOUBLE stacked (44.05%)
57 sims TRIPLE stacked (6.71%)
64 sims QUAD stacked (7.53%)-- This appears to reflect "void" sims. However, on examination, it seems that two void sims are sometimes stacked on two active sims. How much this affects sim performance is unknown.

SUMMARY--TOTAL SIMS SHARING SERVERS WITH ONE OR MORE OTHER SIMS: 495 (58.29%)


THE CONFRONTATION
=================
We approached Linden Lab again regarding this matter.. LindenLab continued to maintain that sims do not share servers under any circumstance. So we presented our data findings. After pressing the matter and speaking to a tech employee, we were told that "server" isn't an accurate term. They do share SERVERS. But each server is 'a double processor server and fast enough to handle the load'.

We presented LL with questions. What kind of "double processor" server? Does it share ram banks? Does it share bus structure, internet bandwidth, HARD DRIVE access? Linden Lab could not come up with immediate answers to these questions (not even the "tech" we were speaking to).

At that time I specifically requested the following information:
* Server brand name, model number and configuration
* Average cost of a server

I was told that this information would be found and they would forward it to me. As of this writing this information has not been provided.

DISCOVERY #2
============
When we searched through the data, we discovered that our sim, ElvenGlen, was sharing a server with EATON. Those who are acquainted with Eaton know that it has been the #1 traffic sim on SL almost since its inception. 24/7 tringo games, massive shopping malls, casinos, nightclubs. We are good friends with the owners of Eaton. Wayfinder designed their kickboxing arena and the game itself. So we know Eaton. It lags like a fiend. And THAT is the sim ElvenGlen had been sharing servers with! As the owner of Eaton humorously said when this information was presented to him: "I wouldn't want to share a server with me."

Further, we made some additional tests. We discovered (from what we can discern) that the RUN TIME ms reading that appears in the statistics box, is apparently not a rating of individual sim content as we had been told by LL... but is instead a rating of total server activity. This went a long way toward explaining why it is that this figure would jump from low to high for no explainable reason. This was most upsetting, because we had been lead to believe (directly told, actually) by LL that Run Tasks ms was the most accurate reading of our specific sim performance. Yet now, we discover that our sim was obviously being directly influenced by the other sim(s) sharing our server-- and that, despite claims to the contrary, what happens on one of those sims can and does directly affect other sims on the same server. (Linden Lab has continued to deny these findings, claiming that their tests prove that doubling up sims on a server does not affect individual sim performance. We have reason to question their data analysis).

We have spent MONTHS trying to trace these problems down. That entire time, LL absolutely denied that sims used shared servers. They kept blaming lag on "user content"... despite our constant data readings indicating that such claims were not correct.

THE DATA
========
As an example...
When we first built ElvenGlen, we had 750 active scripts on our sim-- an amount most people would say is excessive. As widespread as this belief is however, it appears to be an old wives tale. Yes, certain types of active scripts CAN and DO lag sims(as AO devices aptly prove). But standard scripts such as doors, chairs, many types of particles, etc... those are part of everyday SL life. The fact that active scripts are not absolutely responsible for sim lag was established when we visited one sim with 1,500+ *active* scripts... yet boasted an FPS of 3,600 with zero perceivable lag!

WE DECIDE TO ESTABLISH THE FACTS
================================
To prove our point, we reduced active scripts on ElvenGlen from 750 down to about 480. Result: there was NO PERCEIVABLE DIFFERENCE IN SIM SPEED as a result. So we intentionally increased our active scripts by 180... and SIM SPEED ACTUALLY INCREASED over the span of a few days. Conclusion: the scripts on our sim had no discernable effect on overall sim performance. These findings have been verified by other sim owners who have run their own tests and come to similar conclusions.

At one time, a Linden recommended that we try shutting off script operation on the sim. So we did. The results: sim performance did not significantly change. The Linden at that time said, "Hmm.... I'll check into that and get back with you." Despite reminders, we did not hear further on that result.

THE PROBLEM
===========
The fact that we had to run such tests, over and over, based on claims that LL was making-- while they withheld the vital piece of information that SIMS ARE DOUBLE-STACKED ON SERVERS-- did not make us happy campers.

We have been told that server doubling has been discussed on the forums. However, has been estimated on these forums that only 10% of SL users ever visit the forums. We ourselves do not have the time to read every forum post that appears on the web (because of this, a copy of this article is going to be sent by volunteers directly to SL sim owners).

When we have a problem that arises with our sim, we expect straight answers to straight questions. We do not expect to be told things that are not true, nor have vital information withheld. When we are told that "Linden Lab never puts two sims on the same server"... that is what we expect to be the case.

Has Linden Lab deceived its clients, misused payments, provided misinformation? Linden Lab claims that apparently this is a matter of miscommunication/misunderstanding. What they MEANT to say is that two sims are never on the same CPU. (However, we note the above figures, which shows that claim to apparently be incorrect... to the tune of some 51 sims).

We make no claims at this time. We will note, as did David above, that dual-processor servers are NOT the same as two separate servers. We will note that LL recently raised its server setup price from $900 to $1,200-- ostensibly to pay for more expensive, more powerful servers. We note that we were TOLD that two sims never share a server. Those are the facts. We ourselves make no claims, no direct accusations. It is not our position to act as judge and jury. We are presenting the raw data for your examination, as well as recount of our experiences in these matters.

We will say this: Elf Clan did not pay Linden Lab almost $3,000 to set us up on shared servers. We do not pay some $600 a month to operate on shared servers. We were not AWARE that we were running on shared servers and in fact had been told that we were NOT on shared servers, that lag problems were caused by client content and were not server oriented. We were told this over and over again, despite all data to the contrary.

FACT: we can't get 20 avs on a sim without having it lag significantly, nor 40 without it lagging to a standstill. So we have to ask: how is it that a "dual processor server can handle the load of two sims"? This obviously is not the case.

We'll leave this with a comment made during a sim-owner conversation:
"My guess is that Linden Lab simply doesn't have enough private sim owners yet to double-stack all existing servers. And that eventually, they'll reach the 100% point unless it's stopped."

We report these facts and let the association of sim owners make whatever decisions deemed proper.

--o--

* PS-- advance note to LL groupies: We did not present this data to start yet another endless and useless debate with you. The purpose if this post is to inform sim owners of LL activities and the disposition of their sims and funds. We simply present the facts; if some users choose to discard or ignore these facts, that is your decision. We're conducting business here. Were have no desire to waste our time arguing personal perceptions of computer philosophy.

I wouldnt wouldnt have bashed them in the forums, if i was philip and read this i would tell my tech department to stack 10 sims with yours. they have qualified individuals running the show over there and i have full trust in them, I used to own a decent amount of land and still own a sim..sl is constantly growing and upgrading.. so chances are if this is a correct issue it will be fixed soon.. i remember when some of the servers sl ran on were older than others because they were in the process of moving to the new servers...they fixed that quickly and it was great!

Billy, I agree with you that LL does some good things. I agree that SL is hopefully progressing. That is not the issue here. The issue here is sim owners being told one thing and finding out another. That is the bottom line.

One thing I have always credited LL with: they allow open discussion on these forums with little or no heavy-handedness. That is likely one of their #1 good traits. We invite LL response with the same freedom that they allow these forums to exist. We have presented nothing here but factual information. We are not "LL bashing". We are presenting fact, with witnesses to these things. That's all it is.

Like other users, we desire to improve SL. That isn't done by sitting on such information and hoping everything gets better. Frankly Billy, while I appreciate your undying loyalty, if after reading the above post you "have full trust in them"... well, that's your decision. I trust people by their activities. I am likely as enthusiastic or moreso about SL as anyone else. That is not what I *and others* are addressing here. We are addressing client support policies and business practices. Personal relationships with LL are not the issue.

I dont think LL would intentionally mislead residents though, that is where me and you seperate paths as far as agreement goes, if you want ill call on three way with you to LL and ask them about the servers and see if they tell me the same thing they told you.. you know some of the lindens that are in world dont work from LL they are just customer service.. and wouldnt know a server from a can of beans.

I think we need to make note at this time: we didn't come directly to the forums with these matters. We *have* spoken with LL. We did present this information to them prior to posting here. And this isn't the result of just a recent situation: we have been trying to work with LL on these issues for MONTHS. And we were not told these things by just one LL employee, nor by employees who were "just customer service"-- as we indicated in the information above. There is no indication from LL at this point that it plans to correct these matters or change their current business practices. So this leaves us with the hard choices.

We do not take these things lightly, nor act in haste, nor without forethought. We do not post this in malice. We have no desire whatsoever to harm LL. We post this as information for sim owners who are paying thousands of dollars a year for their sims and who have certain expectations out of the performance of those sims.

Let me ask you a question: if your gas company told you they would deliver a certain amount of heating gas for $X and you later discovered that for the last 6 months they had been delivering less than the promised quantity... what do you think you should do? Just forget the facts? Not let anyone else know about these things, just keep it to yourself? If you went to the company week after week and tried to get this corrected and they did not act to your satisfaction (oh, they have their reasons... our new gas is more powerful... LOL)... then what is it your right as a consumer to do?

You go to the papers and you release your findings. That is what we have done.

whom from LL did you speak to about these matters? If you don't want to post it msg me in world.

Just a question. When you were buying your sim, where exactly did it say you would get a dedicated machine? I don't recall seeing that when purchasing mine. Sure, it sucks, but you can't say you were misled.

As far as your testing goes, I'd suggest finding two empty sims sharing the same machine (not void sims) and really lag the heck out of one of them while measuring performance in each with some utility scripts. Post your actual data here. I think that would give a better indication of the impact you're talking about than using two sims that are full of unknown objects and scripts.

I agree with you that it can be frustrating tracking down slowdown issues in a sim, and definitely look forward to the new script processing routines as implemented in 1.7 (coming within the week?).

Perhaps you should-re-read the statements made from sim owners/partners at the beginning of this thread. I would have to believe this is just a small sample of those who have been told that their sims would be on dedicated servers. I was told this directly. And yes, I *can* say we were mislead. This has already been stated and the reasons given.



With all due and earnest respect--I'd suggest that you do so. We've done our research. We've put in the hours and weeks and months. We don't have to keep revisiting the same restaurant to prove to ourselves yet again that they have lousy food... and we have no obligation to "prove" such to others. The data is already presented, both here and in other posts in other places in this forum. We are not the only users to ever question the server-side operation of Linden Lab.

If someone doesn't trust our findings, rather than ask us to post yet more data that some people will for whatever personal reasons not accept, we recommend they spend their own time checking it out for themselves-- and do so in a true, unbiased fashion (ie, do the anlaysis correctly without skewing the results via predisposed bias ). Then the doubter can be satisfied as to the results.

Myself, I'd love to visit LL and take over a server or two in test mode. But you and I both know that's not likely to happen. And frankly, I'm not willing to take the time away from my own business to do a free systems analysis on Linden Lab servers.

I know from my past experience with a sim, that it would come to a crawl when i was the only one on it, in a private sim. However I know that when the sim owner called about it, she was told that it was linked to another sim without her permission. LL got an earful on that one. However I think that LL needs to speak up on these matters. Then again, they can just go and change the TOS at there will and make it all fit to whatever they want, as we have seen in the past. LL created a great world here, it's sad that they seem to say one thing while doing another. I think if they just came out and said, Hey we know you are having trouble, we are working on the problem and we need to mix a few sims together till we can resolve the issue, it would be less of a problem. i find it funny that when i went to both BL's i had absolutely NO lag. Were talking about some of the most complex builds in SL for a short period of time. What servers were those running on.. and why is it that they could dedicate the power to it, yet not to ppl paying for it.
Cmon Phil, give us an answer. As a V.P. of a web hosting company, I know that you can't always control the situations due to technical problems or issues relating to an outside provider, but my company has always been honest to the ppl that pay us for our services.

I am aware that LL stated they are not a web hosting company, but to avoid certain issues pertaining to the DMCA they were a hosting provider. however they are in essence RENTING servers. Not much different than Hosting providers or co-lo's. The money that I put into SL is used to make the content and entertainment for myself and others much more enjoyable. I don't see why a simple request from owners such as myself and others should go unnoticed. All we are really asking for is an answer to what happened, why it happened, and when it will be fixed.

Just my .02

1. "Dedicated CPU" and "dedicated server" are two terms that are easily confused.
   
2. Some time ago, every sim had it's own dedicated hardware.
   
3. There was never a general "attention all users, we are rolling out dual CPU servers now, some sims will be sharing" announcement to my knowledge.
   

So, yes, sims *used to* be on single machines, and that line of thinking has carried over to the present. I still don't believe there's anything in any Linden Lab documentation that any of us reviewed when purchasing our islands that promised us dedicated hardware though.

Like you, I was *also* under the impression that my sim was located on a dedicated machine, and I was *also* distressed when I discovered that it was not. But again, I was never told that it would be.

I know you've paid a lot of money and that you're awfully upset. I am, maybe surprisingly to you, in the same boat. Maybe worse so. I have a sim that will not hold 15 avatars without lagging unbearably. I've never had more than 20 because people leave in frustration *before* that happens. Why is my sim performance poor? Because I have craploads of scripts from other people in there that I refuse to remove. I'm eagerly waiting for the changes in 1.7 because I'm too lazy to review and optimize each script in my sim.

Man, you don't have a clue what your ranting about...

class0, dedicated
class1, dedicated
class2, dedicated
class3 (dualies), 2 sims on 1 server
class4 (dual core dualies), 4 sims on 1 server.

No LL didn't screw you. No I didn't read the whole thing Machines get faster and smaller... where 1 box (class4) CAN run 4 sims at the same time BETTER then a class2 dedicated CPU blah blah blah.

EDIT: Some time ago there were launcher problems, IE the stacked sims, it's not happening as much anymore. It's a simple program error, they aren't doing this on purpose.

Nexus, ol boy, anyone who can read the above post and still call us clueless... LOL. I'm sure you have a clue. Good for you.

Question: Where was that data posted? Some obscure forum post somewhere?

Question: how does that impact direct statements made to sim owners at the time of purchasing sims?

We have accused noone of directly lying to us. Then again, anyone who thinks a person is above a lie just because they're a Linden would be a little naive. Evidence appears to indicate that such did happen from time to time. However, lying is a strong term and I'd prefer not to go there, because that has to do with intent and I try not to judge individuals without indisputable evidence... and that's hard if not impossible to obtain under this situation (ie, did they lie, or were they just amazingly confused?). Without doubt there was at least one occasion when it was just simple miscommunication. Such happens. But it shouldn't happen over and over and over... especially in response to a direct question.

Did anyone ever withhold information from LL clients regarding things they thought might detrimentally reflect on LL? I think the data indicates that's a big YES. I think I would be naive at this point to believe otherwise.

Wayfinder, again, I'm not sure who else at Linden I can forward you to to correct the misgivings you have about us and how we operate, and correct the errant information you continue to bring forth.

Here's what I can contribute.

To the extent of my knowledge and research, we have not stated on the secondlife.com webpages, the Wiki, nor in any announcement or other official posting, that residents own the hardware. Residents are not purchasing, and do not own, the hardware Linden uses for private islands. We do not state this officially, and we don't want the relationship with an island owner to be tainted by the false perspective that we're doing something with their hardware. If something we wrote somewhere says otherwise, please, let me know so we can have it corrected promptly.

So far, the only "evidence" of this that has been brought to my attention are casual references made by Lindens during one-to-one conversations at events and gatherings. So far, each recollection of such an event involved the Linden describing the island ownership as "basically they're buying their own server". If I said that, it would be only as a tool to help grossly simply the concept of what it IS being purchased to someone totally unfamiliar with the process. I would not take such a statement as policy or a contract. I wouldn't even take it the same as if they said they "technically" or "legally" bought the space on the server.

The initial setup fee for a private island reflects (but is not a 1:1 sum of) the costs of setting the island up. This does include the investment made in hardware, the work in setting the hardware up, and the time and effort to create the island and bring it online.

As with anything you buy, what's important is whether or not you feel what you get is worth what you pay for. If anyone's making that purchase decision with the misunderstanding that they get total control over a specific piece of hardware, please, point them my way. I'll do everything I can to make sure we do not convey that image. That's not what we currently provide, and Second Life isn't at the "platform" stage where we can offer that.

Since the Island program began, we've added measures to ensure that private islands always run on the class of hardware they come online with (or better). Prior to this, there was much chaos as private island owners intentionally crashed their regions to compete for better hardware. We then phased out class1 servers entirely, and upgraded class1 private islands, no charge.

Before purchasing any multi-CPU systems, we tested them thoroughly. (They're multi-CPU, Wayfinder, and that is a dramatically difference performance situation than multi-core. If you can please correct this in any future posts, it'd be helpful.) The machines are well over twice as fast as a class2 machine. The CPUs are about 25% faster, and the total memory bandwidth is very, very fast (yes, over twice).

We tried to get one simulator to impact the other's performance. It was one of our primary concerns. We tested it in a controlled situation where we could see and control all variables involved. The two sims just don't affect another. This is what the hardware is designed to do.

Here's what I see today, just off of a rough count:

~235 sims running on older Class2 hardware (1CPU, 1 sim)
~785 sims running on newer Class3 hardware (2CPU, 2 sims)
~180 sims running on brand-new Class4 hardware (4CPU, 4 sims)

I looked up non-void sims in this query. Every last resident-owned region in Second Life are one sim on one CPU. (If that's not the case, it's very wrong. We fix those cases IMMEDIATELY.) Resident sims are not placed on the same hardware as void sims.

Anyone who has actually spent their day fixing situation after situation that each person only describes as "lag", and had their livelihood based on how effectively they can troubleshoot and diagnose each different situation, knows (and this is the one time that I myself will say, "for a fact" that no two people ever mean the same thing. A resident saying "lag" means nothing more than something is not behaving as desired, and I've discussed this in the forum. If someone claims "lag", you are exactly zero steps further to solving the problem; you have eliminated exactly zero potential causes; and in fact, you're probably in trouble if the resident provide real information ("I dunno, it lags. It's slow." Jeffrey Gomez's excellent post in Wayfinder's other "Linden myths" thread covered a handful of problems, from client CPU to other programs to incorrect bandwidth settings to data loss at an ISP to actual problems on the region, that are collectively lumped as "lag" by those who (rightfully) don't care for the technical details.

So, as I've reiterated things beyond anything I can politely call "at length", several Lindens have taken the time to discuss your concerns, and we're still at the point where you point to data you do not share and say that Linden is deliberately deceiving everyone, I can only make a few challenges:

* I'd like to see this data. At the very least, I can forward it to the engineers that tested our hardware and signed off that simulators do not impact each other, and see what does and doesn't match up, and what (if any) flaws in the data-gathering process you (or we) may have come across.

* I note that Elvenglen is currently on the same simulator as the region Seijaku, which (this morning at least) was quite empty. How's your performance today?

* Finally, what action do you seek from us? I'm not sure how I can correct "deception" any more than I've done in this post. Do you want to go back to a single-CPU server (even though it will be slower)? Do you want me to isolate Elvenglen to once and for all separate any performance issues from the presence or absence of another sim? Do you want placed on the even-faster class4 servers, to see if it's worse or better?

We don't have the infrastructure to provide any sort of "upgrade" for everyone, but I'm more than willing to see what we can test, with the regions Forcythia owns (as you continue to speak on her behalf), to once and for all alleviate your fears about our intentions.

Wayfidner has performed a tremendous task and used the formidable resources of the Elf Clan members to reveal and document what many of us suspicioned and inherently 'knew' was the the LL practice of running more than one sim on a server. Kudos to you, Wayfinder, for your efforts.

Wise allocation of resources is a legitimate business practice. Misrepresentation of goods, products and services is the first step towards the forced implemenation of the business plan Exit Strategy.

It seems odd that when an accountability was called for and a lindin finally stood up and gave one, that it was met with such a statement. While I agree that the original post was well done, i felt that the statement made by lee was well written and well explained.. it was what i asked for to clarify things in my own mind. and by no means am i a big LL preacher(pardon the pun). But what the original post called for was accountability and he got it. as well as more. I wish a lindin would do that for me. lol

eek.. now i am getting all teary eye'd in forum politics.
blah
kudos to the plaintif, and the defendant

I'm not jumping on any bandwagon, for or against, but I have a question to the lindens:

While the server software is made to run one per CPU, and is not designed to run on more than one CPU, I can handel that. The question is, is there a bottleneck in any other the following:

Memory.
Hard drive access times.
Network bandwidth from the server.


Is it possible spikes in network performance on one virtual server could be affecting the other virtual servers on the same machine, due to access bottlenecks that have nothing to do with clock-speed?

I'm NOT saying this is the case, nor hinting at it. I am just curious if it might be causing the problems, and if so, is it solvable?

Thank you.
Foolish Frost

Uh, no. There's no trial here. We've explained from day one that the servers with multiple CPUs host one sim per CPU. Every line that has been bandied in these threads as a sign that we've been deceitful, is our attempts to convey that we do NOT place multiple sims on one CPU. THAT scenario does cause bad performance and we do everything we can to prevent it from happening.

Yes, servers with multiple CPUs host multiple sims. We deliberately purchased servers that could run multiple sims with each sim getting better performance than before, no matter what normal activity occurs on either or both sims, without one sim affecting the other. That is what we provide. Island owners get better performance out of it, and the sister sim does not affect them. Everyone wins unless you insist on inventing ways to lose.

Everyone who's ever asked has been told the truth. Every statement we've made has been honest and forthcoming. There is no revelation being made here. This is all nothing new for those who have listened.

If you feel we're being deceitful, I know several better-informed private island owners who will gladly trade their class2 slot for your better class3...

Foolish: It's good of you to ask. It was extensively tested before we bought the hardware. It's been monitored ever since. Conflicts simply don't happen with normal (even heavy) sim usage.

Elror: I'm still looking for any examples of misrepresentation. Feel free to forward any case where a Linden post or document indicates the private island purchase specifically includes a server.

And again, if someone wishes to revert to the class-2 hardware... if a single CPU is more important than the actual performance you get... I think you'll find a few people willing to swap troubles with you.

I was being facetious, Lee. It was more of a way to end a portion of a thread that may have had merit, but was starting to look like a flame war. your post finally explained to me what i have been trying to understand about the sims and how they operate since i started with SL. I am glad you replied, and have always wanted to understand more about how they work. I like to see if there are ways to push the envelope so to speak with them, and without that knowledge i never know if it's a script i wrote or the random sayings of different ppl that it was an attached or stacked sim.

Sorry that you felt i was sounding more like the announcer on peoples court than an SLer.

Lee, I want to first of all thank you for responding to my post in a responsible and professional manner. I have always had to give you credit for restrained and thoughtful posts in the forums-- often in the face of fire.

I agree with many of the things you stated in your post. However, there are some that I do not agree with. You state that we continue to bring forth "errant information". We are merely relaying what we have been told by LL... and presenting data that we have accumulated. Nothing more, nothing less. If our information is errant... we are not the source of that errancy.



Such statements have been made at other places than "events and gatherings". Such claims have been made to clients who were considering buying sims. To the point that I have myself, with my own "ears" (eyes, actually), seen statements by other sim owners in which they have said, "I'm tempted to have LL send me my server and pack up and go home." It is a widely believed statement Lee.

I myself, with my own eyes, saw a post in this forum in which LL stated (if my memory serves me) that they were increasing prices of sims from $900 to $1200 to "pay for faster servers" so that the client can have the best equipment available on the market at the time. That seems to indicate that the clients are paying for equipment. It should therefore not be surprising if some clients are under that impression.




I actually don't think that's the primary issue. I think the primary issue is sim owners being told that they will receive dedicated servers when in fact, what they receive are dedicated CPUs. And as Foolish Frost has asked (and I notedly asked last week but received no reply)...

* Do those CPUs share bus lines?
* Do they share a common hard drive?
* Do they share common network access?
* Is there any conceivable time when the two sims would "butt heads" and lag, such as when a large number of avs are on both sims and demanding access to shared resources?

I know LL has stated that each CPU has dedicated RAM. Does that RAM share a common bus? Access the same hard drive? These are simple questions that we have asked LL several times (and in fact have been asked by other users on these forums)... and are not answered.



All that I've ever stated in our posts is quoting what we were told by LL representatives, and that information ranged all the way from "dedicated servers" to "dual core" to "dual CPU" to "dual processor"... with no one seeming to have the exact information. We are aware of the difference between dual core and dual processor. In our quotes, we used the specific terminology the rep used at the time. We do try not to misrepresent LL statements. That would be biased an contrary to our purpose... to assist both users and LL in arriving at a workable solution.



Glad to hear that. Such would of course be the logical steps before establishing such a system. So why does your data conflict with ours? That is a question that bears further study. I'm not doubting your word in your statement Lee... not even a little. I am curious as to data tests that were run, stress factors that were applied, etc etc.




There are indeed many causes of lag and many different effects. But I believe that in general, lag is categorized in several recognized areas and is definable:

1) Trying to access an inventory item and having the inventory system fail to respond with expected speed (and as a result, bringing avatar activities to an absolute standstill).
2) Walking along and suddenly experiencing the "molasses" syndrome... where everyone begins lagging inexplicably
3) Seeing chat messages fail to go through with expected speed.

But really, we're not discussing lag here, are we? That has been discussed to death in another forum. What we are discussing is proper communication with clients and providing what we have paid for. That is the question at hand, not the cause and effect of lag, which has been already discussed to death on these forums to the point that further such discussion regarding lag seems fairly pointless.



Well, if you'd like to see this data, perhaps someone at Linden Lab has collected our dozens of past emails in which we have sent you tables and charts and screen captures all detailing that information.

Because here's what happened Lee... after 4 months of trying to deal with LL regarding that data and getting nowhere, and having that data cluttering up our computers and desktops, we finally threw up our hands in exasperation and deleted it. Sad, I know, but that's how it is. We do have witnesses to the fact that this data existed, how it was conducted, and the effects that it had on our sims. Beyond that, I really have no desire to present any more data. Sorry, but frankly, been there, done that, tired of such being sent to LL and apparently disappearing into limbo.

However, some of that data was presented on a forum here titled "Linden Lab Lag Myths". I think there is enough there of scraps of data to give folks an idea of what we have been discussing. I think it should also be realized that we are not the ONLY users who have compiled such data... nor the only users who have stated--in the forums-- that their findings point to LL server side issues.

I think we also don't have to point out that just about the same time that forum post was begun (or actually shortly thereafter) serious server-side problems were located and admitted-- noteably in inventory and email areas. So I don't think Lee, to be totally honest, that this is our imaginations here-- nor that Linden Lab has not been provided with all the data necessary for your engineers to examine these issues. That's aside from the current issue however-- that of what we have been told regarding server disposition.



I haven't had time to examine ElvenGlen activity today (I don't seem to have as much time to track server stats as I used to-- day after day after day). I know as of yesterday the sim was running just fine... as would be expected when joined to an extremely low activity sim. But again... are we discussing performance in this thread-- or LL customer service and client interactions? The later is the subject. The former has been discussed to death elsewhere.



The point here Lee... is that it took this post to handle the "deception". Ideally, we should have had a problem, went to Linden Lab, had it figured out, we would have learned LL stacks servers, we would have been respected enough to be provided answers to our questions along with any data that we required, and we could have handled this like professional businessmen. That it had to be dragged to these forums before we received acceptable response from Linden Lab is a great part of the issue. Why do we have to keep dragging these things to the forums to get decent response from Linden Lab?



Lee, Forcythia and I both have always had no other goal but to help SL improve. We would have gladly taken part in any test and research project LL would have requested from us. I'm a systems analyst...and I charge a great deal for professional analysis. But I'd have offered my services freely if we'd have just had an ounce of indication from LL that it cared about our observations. And when we contacted LL time and time again with data findings, only to have those findings ignored (or disappear)... we ceased that willingness to assist LL any further.

We know you can't upgrade everyone. As far as your question regarding "do we want to be put on a single slower server"... how can I answer that? I don't have the raw data to make such a decision. Asked basically if we want to return to a 2.8 server rather than a 4.0 server-- we really don't have the basic information necessary to make such a decision.

I know that a dual core, single-sim server would be expected to work far faster than a single core server running at the same speed. Does a 4.0 split-attention server work faster than a 2.8 dedicated server? In relation to Linden Lab, I have no way of knowing.

What we want-- is more client support from LL and greater respect for sim owners. Obviously from the comments in not only this post-- but other posts in these forums, a lot of folks think they're being snowed by LL.

And right or wrong, factual or non-factual-- that's a perception that no company wants. Because if it's factual, the company needs to clean up its act. And if it's not factual, the company needs to determine what is being done that gives that impression.

If a woman stands on a street corner at 10pm wearing a red miniskirt... people are going to draw conclusions, no matter why she's standing there.

But in this case, I don't think we're totally without factual data or reason to support our current claims. The data is there. The experience is there, the research is there. What more can I say?

pageus: Don't worry about it. I'm just still in awe of this entire discussion. I'm not sure why there's applause for "revealing... misinformation" specifically in response to a question I've answered the exact same way since the first day I started working at the office.

Lee, this is part of the user disrespect and smoke blowing that I am discussing.

Yes, Forcythia owns all of our lands. That is not our doing, that is Linden Lab's doing. I have owned land and even a sim in the past, but Linden lab accounting and sim management tools make it extremely difficult for one person to manage our sims as we have them set up and handle the riggors of sim bookkeeping as well. And we had such trouble trying to manage both of our lands under the current LL land tools/tiers/bugs/problems that we decided to combine all lands... many of which I myself paid for.

When we first opened our sims, you yourself told us two things:
1) Only one person can actually own a sim. Partnership or corporate ownership is not possible.
2) Both of us would have total access to sim management tools.

So we had to come to a conscious decision to turn all sim ownership over to her, and sim management over to me, even though we did not want to do it that way.

The very first day we opened the sim, we had to call you and ask why it was that I did not have access to estate tools-- and that has been a point of contention ever since. And this is exactly what we are discussing-- clients being told things that later prove to be false.

Anyone who is acquainted with our sims knows that while the sims are under Forcythia's name and she indeed signs the checks (and in fact, she does a lot more, as all Elf Clan members are well aware, and bless her for it)... the sims were almost totally built and developed by me. I designed the methods by which we charge for land and pay for the sims, designed the foundation charter of Elf Clan, and Forcythia and I have both worked to encourage people to make their homes on these sims. And if you think I haven't discussed these matters with Forcythia fully, you are mistaken.

So when I speak in these matters Lee, know this: I speak for all three of our sims, regardless of who owns them on your books. And although Forcythia has not spoken up in these forums due to her gentle nature, you can be assured she was as shocked and outraged by all this as I was (well, maybe not quite as outraged. After all, I'm the warrior).

Trust me though, you don't want Forcy to dive in to this. I've seen how she handles griefers and it wasn't pretty. Fun... but not pretty.

Well, to be honest, I understood this CPU/SIM ratio since day one. I lucked across some info about

A> Sim server software is not able to parallel run on more than one server.
B> Some servers have more than one processor on the mother board.

I figured it out by logic.

Now, other people seem to have been misinformed by poor information or assumptions, and THAT is the problem I think we are seeing here. They honestly thought they were getting something else. It's not a matter of false advertising, as much as an issue with communication.

Really Lee. I checked with a few people I knew about it, and they did not know about the CPU vs. Server differance. I think that points to a need to not only inform users, but to make sure they understand what they are being informed about.

Perhaps I'm rambling again. I'll be quiet now.

Odd.. when I posted my Reply earlier this afternoon, it immediately followed Wayfinder's post when I went back to the Forum thread to check it. Interesting, but no biggy.

I agree, a timely and very informative response from Lee.