Stop "fixing" the MW scrap it

[Wolfie]

So if they weren't physically there the concept would not be flawed. Why so?
As I can see it, it only takes player's position and bazaar content as a weightful thing to care about. Please help me getting that into my thick skull =)

MW servers right now have to handle all of this:

Player data (graphical information, actions used, position, and all player information)
Retainer data(graphical information, position, full inventory)
Zone data and zoning
Transactions

A MW server that only has to deal with text has to handle the following:

Retainer data (inventory only)
Transactions

The difference between the two is that the former has to transfer 10-100 times the data between itself and a client compared to the latter. Consequently, the former can only handle a few hundred concurrent users and retainers, while the latter can handle in the tens of thousands. The system is buckling under a few hundred users; there is no way this system is going to work when/if this game becomes more popular and servers start seeing populations in the thousands.

Adding more machines will increase the cost of maintenance, increase downtime and start up time in case of outage, and increase programming complexity because software gets more complex naturally the more machines that it needs to run on. These are scalability and stability issues.

Aside from performance issues, there are usability issues (like organization). The ward names are merely suggestions of where to place a retainer, and in no way enforce any kind of organization; they server almost no purpose. You cannot force retainers to sell only what items their current ward allow, because the vast majority of people sell multiple kinds of items and there are not enough retainers to go around (and if you increase the retainer limit, you run into the issues above with stability and scalability). SE's plan to kick out retainers based on arbitrary rules is mediocre at best. In the best case scenario, you kick out a retainer that has no items for sale and piss someone off because he/she lost their spot in the ward. Worst case scenario, you kick out a retainer that still has items for sale. This plan fails to address what happens when a ward becomes full and no retainer is with an empty inventory.

When you stop rendering the retainers, all organization will be done at the point of sale via the interface. Retainers become more like names for shops, similar to how Amazon.com has user-made shops, as does Neopets -- and they work well. Sellers simply dump their items in the shop (or to use FFXIV nomenclature, retainers) and open the shop (and possibly set up a shop message). Any sort of "browsing" that you think you might lose by removing physical retainers can be reconstructed using a menu interface (i.e. when you browse to the category "weaving goods", you are presented with a list of all shops that carry weaving goods).

An interface to buy directly from retainers in their current form would help usability and organization a lot, yes. But it does nothing to solve the issues of scalability and stability.

In summary: physical retainers present organization and hardware challenges, but everything they do now can be replicated in a menu interface.

[Renshi]

MW servers right now have to handle all of this:

Player data (graphical information, actions used, position, and all player information)
Retainer data(graphical information, position, full inventory)
Zone data and zoning
Transactions

A MW server that only has to deal with text has to handle the following:

Retainer data (inventory only)
Transactions

The difference between the two is that the former has to transfer 10-100 times the data between itself and a client compared to the latter. Consequently, the former can only handle a few hundred concurrent users and retainers, while the latter can handle in the tens of thousands. The system is buckling under a few hundred users; there is no way this system is going to work when/if this game becomes more popular and servers start seeing populations in the thousands.

Adding more machines will increase the cost of maintenance, increase downtime and start up time in case of outage, and increase programming complexity because software gets more complex naturally the more machines that it needs to run on. These are scalability and stability issues.

Aside from performance issues, there are usability issues (like organization). The ward names are merely suggestions of where to place a retainer, and in no way enforce any kind of organization; they server almost no purpose. You cannot force retainers to sell only what items their current ward allow, because the vast majority of people sell multiple kinds of items and there are not enough retainers to go around (and if you increase the retainer limit, you run into the issues above with stability and scalability). SE's plan to kick out retainers based on arbitrary rules is mediocre at best. In the best case scenario, you kick out a retainer that has no items for sale and piss someone off because he/she lost their spot in the ward. Worst case scenario, you kick out a retainer that still has items for sale. This plan fails to address what happens when a ward becomes full and no retainer is with an empty inventory.

When you stop rendering the retainers, all organization will be done at the point of sale via the interface. Retainers become more like names for shops, similar to how Amazon.com has user-made shops, as does Neopets -- and they work well. Sellers simply dump their items in the shop (or to use FFXIV nomenclature, retainers) and open the shop (and possibly set up a shop message). Any sort of "browsing" that you think you might lose by removing physical retainers can be reconstructed using a menu interface (i.e. when you browse to the category "weaving goods", you are presented with a list of all shops that carry weaving goods).

An interface to buy directly from retainers in their current form would help usability and organization a lot, yes. But it does nothing to solve the issues of scalability and stability.

In summary: physical retainers present organization and hardware challenges, but everything they do now can be replicated in a menu interface.

So much truth in that post, thank you for your patience =)
I'd like to note that we don't have any idea on how the Dev team implemented this stuff or whatever. Now we just have to hear something from the Moderators or the new Letter of the Producer, where we'll see how they are going to handle this whole thing.

If you can hear us Yoshi-P, please put a note on the Marketing ingame system for the next Letter!

See ya! Time to craft some =D

[normalforce]

MW servers right now have to handle all of this:

Player data (graphical information, actions used, position, and all player information)
Retainer data(graphical information, position, full inventory)
Zone data and zoning
Transactions

A MW server that only has to deal with text has to handle the following:

Retainer data (inventory only)
Transactions

The difference between the two is that the former has to transfer 10-100 times the data between itself and a client compared to the latter. Consequently, the former can only handle a few hundred concurrent users and retainers, while the latter can handle in the tens of thousands. The system is buckling under a few hundred users; there is no way this system is going to work when/if this game becomes more popular and servers start seeing populations in the thousands.

Adding more machines will increase the cost of maintenance, increase downtime and start up time in case of outage, and increase programming complexity because software gets more complex naturally the more machines that it needs to run on. These are scalability and stability issues.

Aside from performance issues, there are usability issues (like organization). The ward names are merely suggestions of where to place a retainer, and in no way enforce any kind of organization; they server almost no purpose. You cannot force retainers to sell only what items their current ward allow, because the vast majority of people sell multiple kinds of items and there are not enough retainers to go around (and if you increase the retainer limit, you run into the issues above with stability and scalability). SE's plan to kick out retainers based on arbitrary rules is mediocre at best. In the best case scenario, you kick out a retainer that has no items for sale and piss someone off because he/she lost their spot in the ward. Worst case scenario, you kick out a retainer that still has items for sale. This plan fails to address what happens when a ward becomes full and no retainer is with an empty inventory.

When you stop rendering the retainers, all organization will be done at the point of sale via the interface. Retainers become more like names for shops, similar to how Amazon.com has user-made shops, as does Neopets -- and they work well. Sellers simply dump their items in the shop (or to use FFXIV nomenclature, retainers) and open the shop (and possibly set up a shop message). Any sort of "browsing" that you think you might lose by removing physical retainers can be reconstructed using a menu interface (i.e. when you browse to the category "weaving goods", you are presented with a list of all shops that carry weaving goods).

An interface to buy directly from retainers in their current form would help usability and organization a lot, yes. But it does nothing to solve the issues of scalability and stability.

In summary: physical retainers present organization and hardware challenges, but everything they do now can be replicated in a menu interface.

Let me preface this by saying I'm all for a buy it now interface, where no one has to enter the Market Wards to purchase an item. That along with better search functionality and a few other gems floating around in these threads and I would be quite happy with the end results.

There are lots of things that I agree to in this reply. But seen as the Market Wards look like they'll be with as a little longer, I still believe it would be beneficial if they were designed in a smarter manner, possibly by splitting the wards into more sections and capping how many retainers can appear in each section.

What does that save us? The server sends less data. (And the ONLY data it needs to send to each client is the retainer's name and character details of that section. Inventory and all other non-graphical data are unnecessary and only sent when asked for by a player.) Likewise it's only sending pertinent player graphical details to the client as well.

And maybe, we can avoid the Wards filling up.

Done right, each section of the Market Wards becomes not much more crowded than Camp Horizon or even the busiest part of Ul'dah, right in front of the repair NPC. That's assuming the problem with the server is it choking on sending too much data at once.

Even at this game's busiest, I've never seen Ul'dah crash, need to be reset or have as many problems as the Wards. And keeping tracking hundreds of players should be a lot more difficult and processor intensive than static retainers.

[Wolfie]

Let me preface this by saying I'm all for a buy it now interface, where no one has to enter the Market Wards to purchase an item. That along with better search functionality and a few other gems floating around in these threads and I would be quite happy with the end results.

There are lots of things that I agree to in this reply. But seen as the Market Wards look like they'll be with as a little longer, I still believe it would be beneficial if they were designed in a smarter manner, possibly by splitting the wards into more sections and capping how many retainers can appear in each section.

What does that save us? The server sends less data. (And the ONLY data it needs to send to each client is the retainer's name and character details of that section. Inventory and all other non-graphical data are unnecessary and only sent when asked for by a player.) Likewise it's only sending pertinent player graphical details to the client as well.

And maybe, we can avoid the Wards filling up.

Done right, each section of the Market Wards becomes not much more crowded than Camp Horizon or even the busiest part of Ul'dah, right in front of the repair NPC. That's assuming the problem with the server is it choking on sending too much data at once.

Even at this game's busiest, I've never seen Ul'dah crash, need to be reset or have as many problems as the Wards. And keeping tracking hundreds of players should be a lot more difficult and processor intensive than static retainers.

I already replied to this earlier and it does not make a difference.

The server is not trying to send all the data from every ward to a client. If you notice, only the retainers immediately next to your character are loaded, as well as those that are starred for searching. This is one of the ways they're curbing the amount of data processing needed on both the server's end and the client's end. Restructuring the zones so that you create empty space between the retainers provides the exact same result that the current method provides, while simultaneously making the wards even more tedious to go through.

[normalforce]

I already replied to this earlier and it does not make a difference.

The server is not trying to send all the data from every ward to a client. If you notice, only the retainers immediately next to your character are loaded, as well as those that are starred for searching. This is one of the ways they're curbing the amount of data processing needed on both the server's end and the client's end. Restructuring the zones so that you create empty space between the retainers provides the exact same result that the current method provides, while simultaneously making the wards even more tedious to go through.

Never said the server was trying to send all the data from every ward to the client. That's silly, and I wouldn't code it that way.