Sharie Criss

Since nobody is going to be able to say where the money came from to fund Sansar, let me provide some additional information. https://www.crunchbase.com/organization/secondlife/funding_rounds/funding_rounds_list Linden Labs has not had an outside influx of cash since 2006. The money for Sansar can only have come from one place - regular income. What is Linden Lab's source of income? Second Life. They don't have another source. Figure it out people. 1+1=2.

Then provide facts on Sansar's funding. Where is the money coming from?

Genius, money is a resource. They are stealing money from SL to fund Sansar. Where do you think the money is coming from to fund Sansar? The VR fairy?

I said resources. Do you know what resources are? Did you know that money is a resource? If LL cut their expenses on SL, that gives them money to run Sansar which has NO INCOME. Shutting down hundreds / thousands of servers would give LL a lot of cash to use to run Sansar. Prove me wrong. Or just continue to insult because you can't......

The biggest cause of this newer level of horrid performance is that LL reduced performance for regions by shoving more regions on a server so they could give resources to Sansar. This happened about a year ago. You didn't realize that your hard earned money you are spending on SL was getting siphoned off to pay for Sansar, did you?

All the solutions offered only address client side lag. The other problem we have is that LL reduced the resources available to regions - cut them in HALF (probably doubling the number of regions per physical server) about a year ago. It was an overnight immediate drop in performance which could be measured using a Mono scripting torture test in a virtually empty region. LL refuses to admit (or acknowledge) they made a change despite all the performance testing which says otherwise. I strongly suspect that resources were taken from SL to give to Sansar, which really sucks because it appears that we are paying for SL AND LL's pet Sansar project. LL is bringing some new features to SL such as Bakes on Mesh and AniMesh, but to be very frank here, unless the servers have the horsepower to handle it (and they don't) it's just going to make everything worse. Already LL has completely abandoned the concept of grid-wide Experience Keys which would have made that whole system MUCH more useful. It's sad, because it's pushing people away.

Oh really? The fact that overnight the performance of my regions dropped like a stone is not LL's fault? Then who's fault is it? I'm not even talking "full of avatars" performance, but when the sim is empty after a restart. Over the years, I've worked hard to get rid of scripted items that perform badly, ensure that the majority of scripted items that I have control over operate efficiently, etc. My script count has dropped almost in half from where it was at the peak. During the time when my script count was significantly higher, I had script run times of 100% with a fair amount of spare time. My build has not significantly changed outside of converting a lot of old prim and sculpt items to mesh (which increased LI due to how things are calculated), yet now I NEVER have any spare time and my scripts struggle to run at more than 50%. FYI, I DID pay extra for the double prim region yet I'm only using about 15000 land impact now because there is no way it would handle me adding anything more at this point. In other words, I am unable to use the resources I am paying for. My other region which has been pretty much static for years ALSO saw a huge performance drop "overnight." Can you prove that LL did not do something that caused this performance drop? All the evidence I have as well as a LOT of anecdotal evidence from other region owners indicates that SOMETHING changed. What we don't know is WHAT changed. Code? Number of regions per physical server? Hard to say. The sim freezes that occur when avatars TP in and out are ABSOLUTELY LL's fault. Some code is running that causes this behavior. There are MANY bug reports about it (most just get closed as duplicates, and there has been ZERO action on the primary bugs in years. ) Feel free to browse the JIRA. So yes, I'm upset. I'm paying a LOT and things are worse than ever. I think I have a RIGHT to be upset IF what all the evidence appears to indicate is true. The fact that YOUR region isn't having any trouble is moot. Maybe it's mostly empty, has no traffic, hardly has any scripted items, - whatever. Must be nice.

Lol, I made the error of using mixed metaphors, let me fix that.... They MILKED SL for everything they could, NOW they've resorted to bleeding it - which will kill the cow eventually. All in hopes that people will flock to Sansar - the "My Little Pony" edition of SL. Looking over the Bio page for the senior management team, these APPEAR to be bright people - and maybe they are - "Technically." They completely lack a basic understanding however of what their customers want, are completely clueless when it comes to customer service. Anyone remember the days when SL had over 125,000 online residents during busy times? Now we struggle to have 40K, and 20K of those are bots. This is what happens when you fail to understand your customers. You lose them. Having worked in a number of startup environments in my career, been in negotiations with venture capitalists, one of the BIGGEST red flags for a companies lack of success is having a TECHNICAL person as CEO rather than a business person. I strongly believe this is why SL is failing. (edited.) Regarding tier, yes, LL does more, but remember that the current staff is a skeleton crew. Most of the developers are on Sansar. There are lots of other servers and services our tier is paying for - and to be very frank, it's a bit unfair to force landowners to pay for ALL the infrastructure and having "free account" users not owning or renting land basically paying nothing while being able to take advantage of everything us supporters are paying for. All in all, I've accepted that and have been paying the huge tier fees for over 10 years. It's my hobby and hobbies usually have a cost. Until recently, my regions have been performing "okay" - not "great" but "okay." Now they are terrible and I just can't see paying the same rate and getting a LOT less.

Thank you all for kind words of support - we are here because we are passionate about Second Life. It's really frustrating to see the company responsible for the the community we all enjoy so much doing everything in their power to sabotage it. Sure, there is SOME new development, but to be quite honest it almost feels like they are attempting to just toss a steak in the cage every now and then to keep the lions satisfied rather that truly doing things that would REALLY make Second Life a lot better. These are harder problems to solve - severe lag and other performance problems. Based on periodic benchmarks having been run for years, I have pretty solid evidence that Linden Labs has REDUCED server resources allocated to regions thus making performance problems even worse. How much worse? Regions today are running about 50% slower than they were 9 months ago. Considering our Tier hasn't gone down and modern servers are leaps and bounds faster than servers from 10 years ago, with data center, storage, and bandwidth costs ALSO dropping significantly in that time, it's unacceptable to have this performance drop yet still be paying the same. The Pathfinding feature which came out a few years ago could have been Sooo cool to work with - especially coupled with the new Animesh feature. The reality however is that unless your sim is EMPTY, you can't actually USE pathfinding as it performs Soooooo Slowly it's - unusable. Things that used to run smoothly now barely function at all. Second life was designed as a social platform. It was DESIGNED for people to get together, yet regions struggle when more than 20 people actually attempt to be in the same place at the same time. Could Linden Labs fix this? SURE!!!! While not trivial, they could certainly dynamically provide additional CPU / RAM to busy regions to actually HANDLE events gracefully. But - that's a different issue than lack of support. I'll cap this off with the fact that my $10 / month hosting service has 24x7x365 support via phone, chat, email, and support tickets. My SL $400 / month for land plus transaction fees plus premium account fees does not. With how much you are charging me, Linden Labs, this lack of support is outrageous and NOT "normal" in the industry. It feels like you are "Bleeding" your cash cow to pay for your pet project Sansar. There's only one problem - there's no more blood and you've made your cow angry.

Shortly after "Live Support" (which is chat and phone) closed, I had one of my regions crash and not restart. So I did the normal thing and filed a support ticket for offline regions. Up until some point recently (not sure when this policy changed) these types of critical support issues would be addressed no matter what time or day it was. Apparently this is no longer the case. My region was offline until 30 minutes after "live" support re-opened the next day. Chatting with live support, they confirmed that no support at all is offered outside of the chat / phone support hours. Do I need to point out how unfair that is to non-US based customers for this global service that is Second Life? Am I going to be offered any sort of service credit for failure of cloud services? I opened a feature request asking for the ability to Self Service a forced region restart through the land manager, but I anticipate that this will go nowhere. While that would address this particular type of support, there are other types of support (griefing attacks) that will still require Linden Lab to resolve. Linden Labs: This is not the way to treat a paying customer (paying to the tune of over $45,000 USD over a 10 year period.) Do you REALLY want to lose more paying customers? This lack of support coupled with the SEVERE drop in region performance caused either by server consolidation or back-end code changes, it's starting to look like it may be time to find a new hobby. I may just have to take my hobby money and use it elsewhere. I'll be following up on these concerns with a snail mail letter to senior LL management as the LL moderators clearly will NEVER EVER let senior management know that customers MAY be dissatisfied.

Running the 64bit Firestorm may also help. Completely eliminated the thrashing issue for me.

Seriously, that's absolutely silly. The OP asked for a home router, not for a super router to run a large enterprise (where I would use enterprise equipment not some cheapo homebrew box.) You are talking about replacing a sub $100 device that is capable of full wire speed NAT and configured by running a web-based wizard in a matter of minutes with something you have to build for MUCH more cost from bits and parts and mess around with just to get it to work! And no, it's not going to run rings around The ERL that has hardware offloading. A professional would NEVER do this! It's a great hobbyist tip for someone that loves to tinker. I loved to tinker with this kind of thing when I was younger and bored, but now I just want something to just work that is reliable, is commercially supported (security fixes FAST and regular,) has a great community behind it, and doesn't break the bank - and I've got much better things to do with all that time I've saved not constantly messing with my home-brew firewall! This is why a dedicated device built for networking is better than a general purpose CPU / PC (since you seem a little confused): https://help.ubnt.com/hc/en-us/articles/115006567467-EdgeRouter-Hardware-Offloading-Explained As for future upgrade path, um you DO realize that this is a sub $100 device, right? In a few years IF I can get faster than 1G Internet connection at my home I'll just upgrade. I don't think I'll be getting faster than 1G Internet at home though in the next 10 years when most of the world is still stuck on pathetic DSL and Cable services. As for building a box yourself being "always a good idea" I guess we just have to agree to disagree. As a final note, you may want to look at VyOS, it VASTLY outperforms pfsense on the same hardware - like by a factor of 4. Better get tinkering! Maybe in a few weeks you will have it up and running after you've learned how to work with it. Or just buy an ErLite and be running in 20 minutes.

ProTip #2: Ummmmmm For the same money and a LOT less work, the ER-Lite I mentioned also uses about 1/50th of the power and is capable of routing at full gigabit wire speed. I would never use an old PC for this kind of thing anymore - power isn't cheap. A quality router appliance uses a network processor that is DESIGNED to do network routing. Even if your computer is FREE it's costing you more in power to run it as a router in a year than these little boxes cost PLUS their power usage. Again, it's a linux appliance, you get root access, you can hack on it, install (small) apps, etc. I actually swapped out the 2G USB flash drive inside mine with a good quality 32G so I can do more with it If you enjoy tinkering, I can see the desire to DIY, but as a networking professional that does this for a living, I have better things to do with my time

I realize this is a bit old thread and it's probably moot now, but maybe it will help someone else..... From working with so many soho / consumer routers over the years, the conclusion is that most of them are just total junk. Yes, they can speedtest out high, when they aren't busy freezing and crashing, but simple throughput is a garbage number when dealing with gaming (like Second Life.) the difference between web browsing and SL is that SL is continuously creating and destroying THOUSANDS of connections at a high rate. Most consumer routers basically barf on this, being designed for basic web browsing usage. If you are going the consumer route, look for routers that are DESIGNED for gaming. Personally, I'm very fond of the business line from UBNT, like the Edgerouter Lite which is inexpensive and handles SL brilliantly without breaking a sweat. It's Linux based and VERY flexible for people that enjoy getting into the nitty gritty. It does not do wireless. It's JUST a router. For wireless, I add UBNT's Unifi access points which ALSO are business grade, handling a lot of connections unlike most consumer crap. Being separate means a DEDICATED processor just to handle wireless that does not slow down your internet. Integrated cablemodem / routers / wireless is just asking for trouble. Junk. Being business products, they are a bit more oriented towards the IT professional. UBNT has a consumer model which is also good called AmpliFi, that includes optional mesh wireless extenders if needed, and it handles SL nicely too. I know it kinda sounds like an Ad, but UBNT is a NETWORKING company - that's what they do. And they do it well. They don't make mice, keyboards, USB cables, etc. They focus on networking and specialize in carrier grade wireless technology.

The release notes on the wiki for this version show NO changes. What's changing?

Not all sims had objects returned as they had plenty of free prims available. When the LI was recalculated incorrectly, some sims that were near capacity already went over the limit and hence had items returned. This is NOT the first time this EXACT problem happened. It happened in one of my regions back in June of 2015 with the Exact Same Problem. Mesh physics screwed up, LI went wonky, items were returned, and I had to request a rollback. A simple restart, as mentioned in the other discussion, fixes the issue. If you don't have rights to restart the sim yourself, open a support case and ask for a restart.

It not only affected prim counts, it affected the physics of objects. Stairs became blocks, doorways solid walls.... Prim counts were whacked. As I have a lot of spare prims, I didn't have the autoreturn issue that other people had, but a regular restart fixed things for me. If you had things returned, you will NEED to get a rollback of the sim, if not, you probably need to do a restart at the minimum.

The load times and load reliability is absolutely worse, but so is server side lag in general. With the increased prim counts available, there are more scripts, more textures, more mesh than ever before too. Avatar mesh has also exploded with bodies, heads, jewelry, etc. etc. This matters. It means that over the entire grid, there is more the SL servers has to do to deliver a scene to you. A few months back though, something DID change that is causing some things to be significantly worse. When an avatar TP's in and out of a sim, there was always an impact as the servers have a lot of information to transfer server to server. What changed is the level of that impact, causing the sims to completely freeze for 10 seconds or so which is devastating for busy regions. I don't know the cause, only the Lindens know and they are NOT talking about this issue at all. When I asked at one of the office hours meetings, I just got the "There is no update on this issue" response. Maybe LL is trying to save money reducing the physical server count, sticking more regions on a server, etc. We are just guessing here. But it IS a real problem. Regarding speed test results, that data is pretty much completely meaningless. Speed isn't as important as QUALITY of the connection. There are so many factors that go into quality that nobody can tell you why your connection is bad, that's going to take a lot of technical leg work to determine what it is. When you do your testing, pick a test server in Phoenix AZ, where most of the LL servers are located. While they do use content cache servers scattered around the world, your basic connection still goes to LL's primary data centers. Do a VOIP test - this can test the quality of your Internet connection in general Use a hardwired connection - wireless is crap. It's IMPOSSIBLE to get a good solid reliable connection via wireless - I don't care how much you spent on your wireless gear. It's the nature of radio communication and WiFi technology. Use a top quality router - lower end and older hardware doesn't have the horsepower to handle SL well. They were designed for modest needs such as web browsing, that have dozens of simultaneous connections. That's not SL. SL creates MASSIVE numbers of connections to load all the assests, textures, sounds, animations, mesh, object data, physics, etc. etc. It's thousands of connections rather than dozens. There is no speed test that will tell you if your router is capable of properly handling SL or not. Your connection may be throttled by your ISP's equipment too - this may not be detectable by your side at all unless you have access to high end local datacenter servers and bandwidth and can run torture testing accurately. Very very few people have the ability to do this kind of testing. In short, there is no simple answer, but things are subjectively worse in some ways and measurably worse in others from a performance standpoint but the cause is not clear, and LL isn't talking.

Meh. No, really, it's not, because it makes way too many assumptions that are just not realistic in most cases. In fact, when I'm bumping up against the limit, nearly all the items are: NoMod and/or Unrigged. My current look, pretty basic, uses 24 attachment points and NONE can be combined. SL already has a solution for people worried about complexity, and that is the max render complexity limit before it jellies. If you are worried about others using a lot of attachment points, simply lower your render limit.

I have to chime in here because there is some misrepresentation of the facts. While it's true that while running, scripts run in "Spare Time" and won't impact anyone directly, scripted attachments can still significantly impact a sim in many ways. As ChinRey mentioned, scripts can trigger other laggy behavior based on what they do. Worse though, is when scripts are "moved" in and out of a sim. When someone TPs, all the scripts in their attachments need to be suspended and "moved" to the new sim. The code, the data, the current state for each and every script needs to be dealt with. There is the network impact of the actual transfer of that data, but also this triggers the Region change event, which will cause new HTTP connections / handles to be created for any script that needs them, etc. There is also the impact that instantiating the script has on the region's script engine, accepting the data, scheduling, etc. etc. etc. There is a LOT that happens behind the scenes that is completely transparent to you. After all this stuff happens, things settle down and run smoothly again but for heavily loaded avatars it can take several seconds for things to calm down. Meanwhile, everything else in the sim will stutter. This freezing behavior can be seen when testing in an empty sim using an unscripted observer and a heavily scripted test subject, monitoring the region statistics. See what happens when both heavily scripted versus unscripted avatars TP in and out. It may surprise you. Each server only has a fixed amount of real memory (RAM) for each region running on it. Each script takes "some" memory, the more scripts running, the more memory used by those scripts and the less is available to "other" things the sim uses memory for. Once a sim runs out of real memory, it uses virtual memory, which is going to be a crappy desktop-grade SATA drive (cause they are cheap.) This causes the sim to lag horribly as it's swapping in and out bits and parts to and from the disk as needed to that Very Very slow storage. Once a sim starts swapping significantly, you pretty much have to do a restart (which is why I need to restart mine almost daily.) Scripts are not the only cause of low memory here, but they can be a significant factor in a busy sim. There are some venues that will automatically eject people with lots of scripts. What they fail to understand is that by the time they count scripts on someone, it's WAY too late. The damage is done and the sim is already calming down to be only very minimally affected by the avatar's excessive scripts. I would NOT recommend those systems. The bottom line though - if the scripts are in stuff in your sim, go for it. As long as they run smoothly, your good. Just be aware that things may start to behave badly when the laggies hit.

What brand and model is your router? I really would suspect that. I found the majority of consumer routers don't handle the thousands of connections SL makes at a very high rate well at all - especially if they are All In One wireless access point routers.

This has been an issue that has been steadily getting worse over time. There is an open bug (the number escapes me) regarding the lag-spikes and sim freezes being due to mesh attachments. With all the new bodies, heads, etc. especially with Bento, the problem has magnified. No word from LL on the fix or if they are even looking into it. It makes busy sims almost unusable (which is a complete joke in a SOCIAL platform.)

That's absolutely false. Server side lag caused by people TPing in and out is a MAJOR problem for busy sims or anyone relying on scripts / physics. Years and years ago, shortly after Mono was introduced, there was a HUGE problem with Mono scripts that caused everything on the sim to completely freeze for sometimes 15 seconds. It took almost a year to fix that bug, but the problem of attached scripts is still severe. I run a busy sim (30-50 users pretty much all the time) where people usually have a lot of scripts, and you can easily see in Region tools the impact people have coming and going. I have to restart the sim pretty much daily due to the massive memory leaks in the server code. If I don't, it crashes all by itself in 2-3 days, but in Day 2 and beyond, the lag problem is MUCH worse than Day 1. The basic underlying problem is that SL is still based on the archaic 1 virtual CPU per (full) region model, meaning that empty sims are given the same resources as busy sims. The problem is, SL is a SOCIAL platform. This means - people actually get together - on the SAME region. Giving the same resources to EMPTY sims, that are ALWAYS empty as busy sims, where people hang out, makes NO sense. Scripts really need to be pulled out of the mainline CPU and given over to a secondary so they don't drag everything down. Sim owners also need Need NEED a way to limit the maximum CPU impact an individual avatar can cause. Let them run 400 scripts if they want, they will just run SLOW because they are limited to X milliseconds of CPU no matter what. LL fixed the Client Side impact of bad avatars with Jelly Dolls, now they need to fix the Server Side impact of bad avatars.

Support performed a rollback - and the numbers aren't crazy anymore - but the region version is the same.... What the heck?!! Can someone please explain exactly what happened? Am I always going to be at risk now of a region restart trashing my sim?????

OK, I can confirm that it's not JUST a calculation but with alpha masking, regular mesh objects are also impacted. I reverted to blended alphas and the items affected still show crazy LI numbers, so I'm unable to fix this. I need a resolution from LL.