SE's housing system doesn't favor anybody...

[Enla]

,

Except instances don't have to be loaded if they aren't currently in use, which let's face it - most people don't hang out in their houses to begin with. The only time the housing server is ever stressed at present is when new wards get added and everyone is, you guessed it, inside the existing housing instances. After a week it goes back to being a once a year issue and even that can be solved with smarter use of their limited server capacity at present.

[Mwynn]

Except instances don't have to be loaded if they aren't currently in use

Call me crazy but... are we all not playing in Instances such as :

~ Cities
~ Fields
~ Dungeons

It's just a Map that is entirely active all the time (You can see in Field Zones as example when FATEs are still active while nobody is in that "Instance" actually)
"in some cases they're entirely active"

Now for the Dungeon case, it is true that they completely start working / get loaded once you enter them from the beginning.
Basically we do right now play in Instances, hence we had 3x Instances at start of Stormblood, such as having 3x times "The Loch" Zones in their Database.
By choosing 1 ~ 3 we got put in one of the three Copies that were around then, giving us even Triple Hunts and such.

I mean I'm in no Way against any Housing related Suggestions or in this case, giving us more Instances, by doing so tho this Way, or well let's say Square Enix's Way.
We basically had a bazillion Warding Instances that would work just as it would do right now, putting even more load off as they don't de-load if there is nobody around.

(Of course if SE is finally clever in this case, they did it that Way, but knowing SE... they pull something half-a**'d off *eyerolls*)

I see the misuse of the Word "Instance" here basically. Why not apply the Word "Channel" instead, it does the same as it does right now.

[Tridus]

Call me crazy but... are we all not playing in Instances such as :

~ Cities
~ Fields

Those are always active, shared instances. They take server capacity even if nobody is using them, and there are fixed numbers of them. The housing wards are the same thing. Adding more wards means more server load even if zero players are using any of them. These can go over capacity (as we saw in The Lochs at launch) because communication between cilents is an exponential progression: If there's one person in the area, the server has to tell 1 client about the NPCs (Fates, mobs, gatherer nodes, etc). Cost: 1
- If there's 2 people nearby, the server has to tell 2 people what 1 other person is doing in addition to the NPCs. Cost: 4
- If there's 5 people nearby, the server has to tell 5 people what 4 other people are doing in addition to the NPCs. Cost: 25
- If there's 100 people nearby, the server has to tell 100 people what 99 other people are doing in addition to the NPCs. Cost: 10,000
- If there's 200 people nearby, the server has to tell 200 people what 199 other people are doing in addition to the NPCs. Cost: 40,000

That's why the shared instances have an effective cap on how many people can go in. Getting past a certain point will demand massively increasing communications requirements to the clients, which taxes both the server and the clients. But they're also not free even when they're idle, since they have to remain in memory.

Housing interiors and apartments are on demand. They can be turned on and turned off as people enter and leave, meaning a house with nobody inside it isn't using anything except storage (which is comparatively cheap). Dungeons are the same thing. If the server can support 500 dungeon instances at once, it doesn't matter if all 500 are doing Sirensong Sea or if they're in 500 different houses. Although I don't know what SE's backend looks like, in a standard n-tier backend, this part is the easiest to scale. You can effectively open up more capacity for on demand instances by adding another server, provided your backend storage and communications links can handle it. Since it doesn't matter what people are using that capacity for, it gets allocated effeciently based on whatever the playerbase feels like doing at the time, as compared to a shared zone like a ward which is always active even when nobody is using it.

Really, if they're so heavily constrained by server limits, shared wards are not the way to go. Fully meeting demand is going to require at least doubling if not tripling the existing ward count, and that's all dead server capacity a lot of the time.

[NephthysVasudan]

<snip>

The problem with instanced housing that everyone ignores/forgets/overlooks - is when it goes live for first time...your going to have a hell of a time getting things stabilized server wise - then you got storage capacity issues.

Then you got the factor in visitors/public/non-public....when people want to show up.when they don't.
Then the big one.....how many will be there? Available so to speak? There has to be a limit..a cap...it wont be endless.

Multiply that for each-and-every-bloody-server on the network right now....in all worlds....

This exponentially can get out of hand very fast/very quickly. In technology terms it adds up rapidly in terms of cost. - Trust me...I know these things working in a Data-center myself.
In terms of an idea its simple - but in deployment terms and hardware requirements? That's a whole different pandora's box.

And while other games have done this....SE has to deal with a model that's already in operation..already deployed....square peg...meet round hole.
Bigger hammer may not be the appropriate answer..let alone a nuclear bomb.

This is the pro's con's thing I alluded too. housing isn't the same as over-world in terms of functionality and how it behaves.

And for the record - its not a bad idea...it a good idea...but launching said idea....is a bit of a headache.
Which requires a lot of labor...a lot of time...and a lot of debugging....on the level of a full blown patch or more I'd wager based on recent track record.
Can it be done? Yes.
Will it be done? Not likely in the next patch or two...but i could be wrong.

PS: Sorry if i reiterated what you said....sometimes my brain has to play catch up to get what your getting at...my point here...is its not an easy puzzle to sort out...on an already active model.

[Mwynn]

And while other games have done this....SE has to deal with a model that's already in operation..already deployed....square peg...meet round hole.
Bigger hammer may not be the appropriate answer..let alone a nuclear bomb.

Can it be done? Yes.
Will it be done? Not likely in the next patch or two...but i could be wrong.

This is the Point where the Problem lies, if they do something wrong.
Espescially with Housing and we know when it comes to Housing they do something Wrong as usual.

If we take the Korea Version of FFXIV (I believe it was Korea, might be the Chinese Version if I'm wrong now) ...


What they basically did was to merge two Servers into one without thinking of the consequences that both Servers had the same Plot Assignements (Ward 1, Plot 1).
Both Plots owned, they merged, Plots were bugging out, they lost the Plot.

You can imagine how the Community reacted to that, now imagine with ours if something like this or something else happens in regards of the Housing Situation.

One Attempt and a Community that is about to explode about a Issue that is going on for Years.
I assume this will, sadly, not be resolved until in another 5 Years, as the Community tends to grow and the next new Wards will be taken aswell.

[Moonlite]

snip

Funny you should mention the Korea situation. So at least on Balmung when they launched housing one of the first things people asked, "How will this work when they merge servers?" Guess what it turns out the paid professionals didn't ask themselves that question. From the reports it sounded like they asked the Korean company to hold off till they could make the server population irrelevant, like they have done for us.

It is also why they are so dependent on cross server and instance content. They don't have an answer for server merges that doesn't make them look like complete (word of choice here). You could be on the shittiest population server and still be able to get everything done with enough cross server tools.

Does anyone ask why they have limited resources they don't put harder restrictions on purchasing. Sure we would all complain if owning a house required you to do "Payroll IV" or "Bring your A game" or "In the outfield VII". They have numbers on what people are accomplishing it would be easy for them to balance the amount of players who actually get a house.

Once you have the money, you have the opportunity to compete to get the land you want, or attempt to buy it off someone. Thats it. Beyond that, youre footing is the same as everyone else. Thats why the pragmatic choice is to have a few options in various places.

But if people are so dead set on a location, then given time, someone will sell it to you at marked up pricing.

Actually this is considered a violation and a report able offense. You do know that? Inflated price so someone will move is not a currently endorsed move by the talking heads.

[Tridus]

The problem with instanced housing that everyone ignores/forgets/overlooks - is when it goes live for first time...your going to have a hell of a time getting things stabilized server wise

Yep, no way around that.

- then you got storage capacity issues.

Housing interiors are already instanced, so the only extra storage is if they add more of them, which they're doing anyway. I don't see this as a significant problem.

Then you got the factor in visitors/public/non-public....when people want to show up.when they don't.

Apartments already have this. While not perfect, it works reasonably enough. They could also copy what other games have done if they were going to full instanced housing.

Then the big one.....how many will be there? Available so to speak? There has to be a limit..a cap...it wont be endless.

It should be endless. The only acceptable answer at this point is 1 per account per server. Anything else is just hoping to make the group of people left out small enough that they can be ignored.

Multiply that for each-and-every-bloody-server on the network right now....in all worlds....

It'll be a problem initially based on what Raubhan's Wall did to the servers, but once the novelty wears off it shouldn't be. People by and large don't spend tons of time in their houses after the first week. There's no question that it'd be rough early on, but I tend to think people would be more willing to pay that price than be willing to have another "did you login within 1 minute of the servers coming up?" lottery.

This exponentially can get out of hand very fast/very quickly. In technology terms it adds up rapidly in terms of cost. - Trust me...I know these things working in a Data-center myself.
In terms of an idea its simple - but in deployment terms and hardware requirements? That's a whole different pandora's box.

I appreciate the insight of someone on the ops side of the fence, being a developer.

And while other games have done this....SE has to deal with a model that's already in operation..already deployed....square peg...meet round hole.
Bigger hammer may not be the appropriate answer..let alone a nuclear bomb.

Yeah. Layering another system on top of an existing system isn't great... yet it's what they're doing with the glamour wardrobe thing while also keeping the armoire, which kinda does the same thing but only for certain items because reasons.

I mean, adding more wards until the problem goes away is effectively what LotRO did, which wouldn't require adding a new system but would be even more expensive server wise since now you've got all those ward zones AND all the instanced house interiors. That would have to come with an account restriction so someone can't just buy 16 houses on a given server.

Will it be done? Not likely in the next patch or two...but i could be wrong.

I think you're right.

PS: Sorry if i reiterated what you said....sometimes my brain has to play catch up to get what your getting at...my point here...is its not an easy puzzle to sort out...on an already active model.

All good.

(lol, HyoMinPark and I both did an edit incoming post at the same time)

[NephthysVasudan]

<snip>

^Real bonnafide developer....not fake.

Total respect points.

See people make a number of assumptions - and tend to forget the physical side of the coin.
SE has a damn good idea of whats in place...but its never as simple as it seems...they like are using a CLUSTER of servers...multiple servers tied together working in concert....physical..not virtual....and then you got the databases all communicating with each other....and the resources on top of that.

So yes..deceptively simple...but to deploy such a model on an active system? - Welcome to dev hell.

EVE Online faced this challenge and it took them years to get rid of trash code...YEARS. (Over 10-15 years old now)

<3 @Tridus

[Claviusnex]

snip[/B]

Yes it would take some work but sweeping it under the rug and trying to put a band-aid on a hemorrhage just makes matters worse. It will also cost money but I'm not sure that is the issue so much as having to admit that the original idea was not good especially after doubling down to try and force it to work. Even if SE changes course I don't see instanced housing as a patch item but something for a major release.

Part of the development work is determining how much server/storage/network will be needed to support the instances. SE should already have a decent idea as the interiors are already instanced. They also know the exterior requirements that can be added in for a rough total. With this they should be able to start a business case and architecture design. Once the implementation design and code are far enough along they should be able to start performance testing to better fine tune the system requirements and performance code paths. Nothing out of the ordinary as far as the development process goes.

As far as going live nothing says you have to let everyone hit it at once. SE could stage the user base access over some period of time. SE could also hire some folks who are use to working in large scale, high transaction computing to do much of the design and platform development. People who think it terms of 40K-50K of concurrent users plus background batch processing with millions of transactions/second as is done in large commercial systems. This would increase the odds of having capacity requirements determined correctly and working the first time out. It should also help with being able to dynamically add and remove server resources when peak load changes as it is common problem solved regularly on commercial systems. Of course you don't get these types of folks on the cheap.

Again keep in mind instanced housing doesn't have to be an all or nothing proposition. In fact I think it would be best to add it in parallel to the existing ward based system. The key is it makes housing with its benefits available to all who would want it.

[NephthysVasudan]

<snip>

That's all good and fine...but your assuming (without knowing it) that the platform is a start from scratch system.
Don't work that way...because they have to impliment the changes into an already active system...that likely has "crap" all over the place.

EVE Online faced this issue..they fix one major issue - and broke 10 other issues that they didn't even see coming or predicted...and it chain reaction'd from there.
Do I know what SE has under the hood? No.
It's called - the bigger the game - the more complex the code - the more complex the code - the messier it gets - and its an endless cycle.

Am I saying it can't/won't be done? No.
For all we know they got something in the works.
What I am saying is...don't expect this to be over night...let alone next couple of patches.