Month: November 2008

Script capping

Posted on by Inara Pey

During an Office Hours discussion, Babbage Linden laid out what appears to be the future for managing scripts in SL. From this and other discussions, it is evident that the issue has been causing concern within Linden Lab for some time now, and was likely to have been the kick-off point for the recent OpenSpace sim debacle – even if they opted to “leverage” that situation for different purposes…

In a nutshell, and bearing in mind I’m not a techie – the problem is that there is obviously finite memory available on each server that runs SL sims (4 sims per server in the case of “full” island sims; 16 in the case of the current OpenSpace sims). This means that each sim gets around 800Mb of memory in which to run residents’ scripts. Up until now, if there are more scripts on a sim than available memory, the sim has “swapped” memory with the other sims on the server in order to meet the needs of all the scripts it has running.

While this hasn’t been a “major” problem in the past, it has impacted performance (lag). However, because we are all natural script hogs (in the same way we are all prim hogs), the memory swapping situation is approaching epidemic proportions and seriously impacting overall grid performance. Thus, it is fairly evident something must be done.

Linden Lab’s preferred approach to solving this problem is to allocate memory space to sims in much the same way as prims are currently allocated: a finite amount of memory is “given” with the sim, and (like prims) this is divided across the sim in accordance with how the sim is parcelled. Therefore, if you own a sim and divide it into two parcels, each parcel gets roughly 50% of the memory space available to the sim. Divide the sim into 4 parcels, and each parcel gets roughly 1/4 of the memory allocation…and so on.

I say roughly, because at the same time, a “resource pool” on each sim is set aside for Avatars that will enable all the myraid of in-world attachments we all tend to carry around to work.

On the surface, both of these apporaches do actually seem reasonable, although it has to be said – as in the comments given by those quoted in the extracts published by Dari – that there are potentially better solutions. Further, this whole idea raises a plethora of questions, many of which LL have yet to answer – of which by far the most important is how will user awareness be handled?

Right now, we all expect that, no matter what we purchase, when we get it home, it will work. OK, so sometimes it may run a little slowly, but it will work. With the new limitations, this will no longer be the case…if you’re on a 1024 sq m parcel and your new left-hand widget wongler takes you over your resource allocation it won’t work until you sort out which items you currently have rezzed must be returned to your Inventory in order to “make room” for the widget wongler….

It gets worse. Prims are relatively easy to understand. They are finite units, irrespective of their relative size. Thus a prim is a prim whether it is 0.5 x 0.5 x 0.5 or 10 x10 10. Scripts (in terms of memory use are not. Some scripts run in a few Kb of memory space, other require the (current) maximum of 64Kb (thus gobbling more memory). Therefore, thus judging whether the next purchase will work at home or require other things get “put away” to “make room” for it is not simply a matter of counting the number of scripts it contains…

And what about those already exceeding the (currently unspecified) limits on resources? The day before the caps are implemented, they’ll see everything running fine. The day after, there is a good chance they’ll find things are not working at all, leading to a potential outcry as beloved household items suddenly have to returned to Inventory or swapped back and forth with other items, thus making a mockery of having a “home” (when was the last time you had to put the sofa into storage to make way for the television set in rl?).

Unless communications prior to the event are handled clearly and passed to ALL members of the SL community, there is a danger that those getting lambasted for perceived failures with objects will not be Linden Lab….but the creators of the objects themselves. This would be grossly unfair on content creators.

There are many other questions that need to be addressed on the issue – how will protected land work for example (particularly where it is protected to provide prim bonuses to other parcels – will the memory allocation be available to provide a similar “memory bonus”); how will malls work, if memory is not only tied to land size, but land ownership?

Then there is the potential impact of the “resource pool” for Avatars. Again, at the moment it makes no difference memory-wise as to how many agents (Avatars) are on a sim – the CPU will simply start memory-swapping to accommodate until the nominal Agent Limit is reached. Practically this does mean that where the limit is being approached, things get bad on the sim in terms of lag (and on the other 4 sims sharing the server for than matter).

While a resource pool may prevent the latter problem of the other 3 sims on a server being impacted, the fact that resources available to Avatars will be capped means that the numbers of Avatars able to enter popular venues in future could be seriously impacted: get a dozen Avatars on-sim blinged up to their eyeballs, all wearing (God-awful) “clickety” shoes and scripted hair with Chimeras attached to heads / shoulders / butts, as well as two or three other “heavy” attachments – and one can see a sim’s “pool” of Avatar resources quickly drying up, with the result that while others may be able to Tp into a sim, they’ll find that “nothing works” when they get there.

No-one can doubt that when we have finite resources available to all of us, then something must be done to ensure those resources are fairly managed and distributed – and that the reasons for such controls, how they will work and what it will mean to SL users is clearly communicated. Further, it is also necessary to ensure that reasonable tools / facilities are put in place to enable people to husband their available resources wisely.

But given LL’s inability to keep its left hand informed about what its right hand is doing, let alone actually communcate with its user base clearly, concisely and openly, can we really believe they’ll shoulder the responsibilities inherent in introducing such fundamental changes to SL wisely or constructively?

If things do go in this direction, then I, for one, will not be holding my breath.

Farewell, Ohiopehhla

Posted on by Inara Pey

Those who read this blog regularly (assuming people do!), can’t help but notice how proud I’ve been about my home – Ohiopehhla.

Yes, it’s a little egotistical talking about something one has largely made oneself, but the fact remains I am proud of the place, all those we’ve shared it with seem to have been equally happy.

But now it is all at an end. If this had come about through personal circumstances, then letting go would be somewhat easier. However, while we would have been content to continue on the island indefinitely, matters have been taken out of our hands.

Ohiopehhla is an OpenSpace sim. That means that from Jan 5th next year Linden Lab expects us to either a) continue to pay $75 USD a month and see the sim reduced to a 750-prim limit, zero scripts and a “10 avatar” limit, or b) have to pay an initial $95 a month to maintain the sim at the current prim count, but with as-yet unspecified script and use limits imposed upon it, with this amount rising to $125 USD from July 1st 2009.

Neither of these options is either practical or acceptable. I’m not going to rake over the whole OpenSpace sim argument;suffice it to say that Ohiopehhla is now to be abandoned.

Did we “abuse” the sim? In Kingdon’s and Jack Linden’s eyes, the answer is probably “yes”, despite the fact prim usage was kept to between 50-60% of the allowable maximum on an OpenSpace sim and – more importantly – I worked hard to manage scripts and keep the sim performance within the (vague) parameters set for it by using the (primitive) tools Linden Lab provide for managing estates.

We’ll no longer invest in land in SecondLife beyond renting. True, this still means Linden Lab gets tier from us (via our Estate Owner) – so Kingdon’s bottom line is protected and he’ll not lose any sleep. But for us it means creativity has now gone from SL – and we’re not alone in this; many feel that trying to create anything worthwhile in world is now wasted effort, as sooner or later Linden Lab will manage to find a way of knocking it down.

And that is perhaps the greatest tragedy of the entire OpenSpace fiasco.

I’m amazed….

Posted on by Inara Pey

…that there are people willing to accept this ploy.

It’s probably fair to say LL have played somewhat fast and loose and got precisely what they wanted. Forget the hype over the original 66% tier hike – that was never going to happen in one bite, and they knew it.

So, bravo, Mark et al! You’ve achieved what you set out to do, and what I suspect you wanted to do from the start – introduced yet another tier level, a “new” form of sim, & got residents to readily accept it. In doing so, not only have you managed to portray yourselves as “listening” and “responding” – you’ve actually got people congratulating you on the “new policy”!

It’s been a very clever game plan. No doubt this outcome is more-or-less what you wanted from the outset. However, you knew that if Jack simply annouced this in his blog entry of October 28th, the backlash would be huge, and you’d have been left with nowhere to turn or alternatives to offer….

So instead, you cleverly got people’s blood up by overstating the intended price-hike on existing OS sims. Rather than 15%, you went for an improbable 66% – and then you left it up to residents to suggest what you were hoping they would – the introduction of an additional sim type with reduced capabilities.

No wonder Jack so quickly leapt on this during his “impromptu” Conceierge meeting a couple of days after the 28th October announcement.

Thus, by clever baiting and slight-of-hand, you lured people into a backlash against non-existent massive hikes in tier rates, leaving you free to play things out for a week or so before putting forward your intended changes by way of a “compromise” in the knowledge that people would a) come to you in thanks, b) go away believing they had “won” over the issue.

BUT – the “victory” may yet be pyrrhic. OpenLife is maturing and is gaining backing (and not just from disaffected SL users). The pricing policy over there for both “Mainland” and “private clusters” is attractive and is set to allow any increases that might be required in the future without hitting people in the pocket. Yes, it has got some way to go before it can offer all the positive aspects of SL – but they are coming. What is important is that you’ve already caused a lot of talent from SL to make the cross over and start work there.

You may yet find that, as OpenLife (in particular and its sister open source grids in general) continue to rapidly mature, you have through these games, opened the floodgates and will be unable to stem the flow of users out of SL – and with growing competition, can you be so sure the influx of “new blood” you hope the Big Spaceship deal will bring, will compensate?

OS sims: numbers and questions

Posted on by Inara Pey

I read an interview with Mark Kingdon (aka M Linden – has he mistaken Linden Lab for Britain’s MI6, or has he an innocent love of all things Bond?) with Wagner James Au (known in-world as Hamlet Au) on the OS sim situation that raised some interesting questions – and which lead me to do a little research and some math (which I’ve been able to verify as being reasonably correct thanks to Vint Falken’s excellent work). The interview also lead me to again question where Kingdon actually see the future of Second Life residing….

The relevant quote on the numbers front came early on in James’ piece, quote:

“While Kingdon won’t say what exact percent of the world is now comprised of Openspaces (“That’s not a number we publish”), he acknowledges it’s been a significant portion of SL’s geographic growth.”

Coupled with comments published elsewhere by James on the farago following the original hike announcement, I started thinking about sim numbers, percentages and suchlike.

At a recent Virtual Worlds conference in London, Kingdon and CFO John Zdanowski (or “Z” – another with a Bond fetishist?) intimated that there are around 32,000 sims deployed on the SL grid, of which we know some 5082 are Mainland, and supported figures put the number of Openspace sims at around 13,000.

By my reckoning that makes just under 42% of SL “OpenSpace”, which kind of makes Kingdon’s assertion that “, “this price adjustment affects only a portion of land in Second Life;” something of a stretch.

It also reveals something far more interesting of note:

  • IF the hike went through at 66%, a loss of around 40% of these OpenSpace sims (5200) could be absorbed by LL without significant financial hardship
  • Some 1300 servers would be saved out by LL, ready to be recycled back into the grid, the majority of their original cost already covered
  • SL would suffer a land mass reduction of some 20% in the process (which might be viewed as a good trade-off by LL, given the amount of Mainland lying fallow).

Now, were we to take these figures a little further and adjust them ever so slightly, the trend remains favourable.

For example: if the tier hike is dropped to 30%, LL could still absorb around an 18% loss in used OS sims before feeling the pinch, and see around 600 servers saved out for a more manageable reduction in overall landmass (under 10% lost).

So… could this be the real art of LL’s game?

Will we see that tomorrow’s announcement will offer a “compromise” wherein LL will still up tiers, but at a “reduced” rate of no more than 30%? After all the furore over the original announcement, this could certainly be spun as a demonstration of LL’s “willingness” to “respond” to resident concerns – and yet get what they original set out to achieve: increased revenues from the surviving OS sims while cycling servers back for re-use and in part redressing land use towards the Mainland?

Others have already speculated that LL’s game had always been to bring in the tier hike in at a figure somewhat lower than the original 66%, and thanks to Vint’s research and my suggested figures above, it’s pretty evident that if this is indeed the game plan, it has some pretty positive results for LL all around.

There is something else within Kingdon’s comments during his chat with James that caused eyebrows to rise, quote:

“Somewhat related to that, he spoke to a point mentioned in my GigaOM article about the protest, which described the steady decline of Premium subscribers as a subject of great concern for the company. According to him, my inference there isn’t accurate: “That’s one of the only figures that’s going negative,” he said. “You can’t use that as an indicator of the overall health of the land market… Premium subscriptions are immaterial in our overall business.””

…TILT!!!!….”Premium subscriptions are immaterial in our overall business….”

Now…either Mr. Kingdon received a sharp whack to the back of the head when saying this (and as such should a) be forgiven for such an utterance, and b) needs to avoid people standing behind him when speaking in the future) – or he has not just let the cat out of the bag – he’s also given it a first class plane ticket to a destination of it’s own choosing.

As pointed out by the redoubtable Anne O’Toole, if this really is the case – that Premium Account holders are “immaterial”, then it could be extension be argued that Mainland usage is “immaterial in our overall business”, as one of the “perks” to Premium Accounts is access to Mainland parcels at reduced tier. Similarly, given the higher level of support promoted by LL for Premium Account holders, it could be said that technical support is immaterial to LL (actually, some would likely argue this is very much the case anyway, but you get the picture).

And if Premium account holders are “immaterial in our business”, where does that leave the vast number of “free” account holders…?

One could be generous to Kingdon (as Anne herself is) and point out this is an “innocent” (if massively unfortunate) slip of the tongue that simply needs to be redressed.

But “slips of the tongue” like this are more usually the result of the brain accidentally revealing the real thought processes, rather than just getting “words muddled”.

As such, and taken alongside the recent corporate annoucements I’ve previously mentioned, one can’t help but feel Mr. Kingdon has, in this “slip” again indicated he believes the “enterprise solutions” market is where he wishes to pitch the grid and SL, and the rest of us are little more than deadweight Linden Lab has to lug around.

LL’s “future vision”

Posted on by Inara Pey

Is it any coincidence that on the 20th October LL signed a deal with Rivers Run Red to actively promote “immersive services”, described with inevitable fanfare and hype on the 21st October was followed at the end of the week by the announcement of the OpenSim tier hike?

It is pretty evident that LL view their strategic future is with “enterprise solutions”, despite the dismal (albeit premature) attempts to move the grid in that direction last year.

As such, it could be suggested that the OS sim tier hike is all part of this “corporate” strategy. Despite all the negativity surrounding it right now, LL know that while many may well vote with their feet and leave for places like OpenLifeGrid – there will be many more who will essentially “suck it up and live with it” – and these people are liable to be far more compliant when the next round of “changes” comes into force.

And the loss of OS sims could even be viewed in some circles within LL as “beneficial”:

  • If there are technical concerns relating to the management of said sims, the fact that many are already being returned to LL allieviates these concerns more easily than changing the OS specs….
  • The people “leaving” SL can be viewed as “troublemakers” who are best gone from the grid
  • Counter-intuitively, a drop in resident numbers = a drop in daily concurrent log-ins = a “more stable SL expereince” (for one thing the Asset Cluster won’t be as wobbly)
  • It leaves LL free to “adjust” things elsewhere to make Mainland more competitive with full private islands / estates
  • It leaves LL free to *possibly* – as Anne O’Toole has suggested – move into the “fully prefab sim” market: offer NEW residents a nice OS-like sim, complete with a choice of landscapes and house directly from LL themselves, at a nice, tidy price & reasonable tier

And of course, lastly, it helps reduce the overheads on managing SL so they can be focused on the wider aspects of the grid and so-called “corporate solutions”.

Does this sound overly cynical or naive on my part? Possibly, but given the drive to make the grid a “corporate tool”, one cannot help but wonder…