Denaut
Trump's Staff
- 2,739
- 1,279
Found this really goodtalk from a Zynga engineer about scalability at GDC. It is just the right amount of technical that it is thorough but understandable. Worth watching if anyone here is interested.
Pan'Theon: Rise' of th'e Fal'Len - #1 Thread in MMO
- Thread starter gogojira_sl
- Start date
Reading up on Spanner hurts my brain. If I'm reading correctly, that's not Google's primary database, it's a project they have using some very smart people focusing on cross datacenter replication for google features. Bigtable if I understand correctly is their logical volume management. These are not really relevant things for anyone who doesn't have google's top tier engineers focused on their particular proprietary high end DB needs.
Just to add, it appears google moved from MYSQL to MARIADB:
Google Waves Goodbye To MySQL In Favor Of MariaDB ReadWrite
The reason Facebook used HBASE makes sense. They could throw thousands upon thousands of servers at a problem. It's the same reason a small MMO startup really shouldn't, in it's cost prohibitive if you don't have Facebooks Billions of dollars:
The Apache Hadoop software library is a framework that allows for the distributed processing of large data sets across clusters of computers using simple programming models. It is designed to scale up from single servers to thousands of machines, each offering local computation and storage. Rather than rely on hardware to deliver high-availability, the library itself is designed to detect and handle failures at the application layer, so delivering a highly-available service on top of a cluster of computers, each of which may be prone to failures.
Also, why would they use Hadoop for their messages but mySQL for the other 90% of their data? I'm guessing there is some app reason.
Again, the base point is for any sort of transaction that requires data access, be it select or alter you still need a plan in place for tuning unless you have unlimited resources for servers. NOSQL or relational you still need to tune.
Just to add, it appears google moved from MYSQL to MARIADB:
Google Waves Goodbye To MySQL In Favor Of MariaDB ReadWrite
The reason Facebook used HBASE makes sense. They could throw thousands upon thousands of servers at a problem. It's the same reason a small MMO startup really shouldn't, in it's cost prohibitive if you don't have Facebooks Billions of dollars:
The Apache Hadoop software library is a framework that allows for the distributed processing of large data sets across clusters of computers using simple programming models. It is designed to scale up from single servers to thousands of machines, each offering local computation and storage. Rather than rely on hardware to deliver high-availability, the library itself is designed to detect and handle failures at the application layer, so delivering a highly-available service on top of a cluster of computers, each of which may be prone to failures.
Also, why would they use Hadoop for their messages but mySQL for the other 90% of their data? I'm guessing there is some app reason.
Again, the base point is for any sort of transaction that requires data access, be it select or alter you still need a plan in place for tuning unless you have unlimited resources for servers. NOSQL or relational you still need to tune.
So opinions on if the Kickstarter fails to fund , should Brad immediately go for self funding through Pantheon website , or wait a bit and present it better on his own site or another kickstarter ?
Whenthis KS fails they are required to re-read this thread a dozen times, then start over.
- 7,547
- 11,830
I think they should reorganize by spending a month planning a Kickstarter for less money that is more clearly to produce a tech demo, get infrastructure in place, hire a business manager and get set up to the point they can legitimately have a KS or other crowdfunding for the GAME that will then not feel like such a sham for them to promote and pitch.
edit to add: I think this is the only way I'd keep supporting the project. By them saying, okay, we took too big an initial step, let's get back to basics and start right, even if it's a slower start. I'm not just going to go throw them money on a website though for the same thing we aren't getting from the KS.
Denaut
Trump's Staff
- 2,739
- 1,279
That kind of low level tuning is done by our server programmers writing server code on a per project (or per module rather since projects share modules) basis. We don't need a dedicated DB Admin because the programmers themselves are in a better position to optimize in such a setup.
If you are interested in the technical back-end on online MMORPGs, it is very interesting stuff and only extremely recently has some of this very new technology started coming into fashion. Mostly because big companies like Facebook and Google started using and experimenting with it, it gained some traction. For those companies it doesn't fulfill all of their needs, they are huge companies with all sorts of different services, so those kinds of companies use different techniques in different places.
For applications and services where ACID is not important, but latency, read-write speed, and scalability are important NoSQL with functional concurrent programming languages are incredibly powerful. Facebook's chat service (which uses Erlang as well) is one example, and MMORPGs are another one.
Funnily enough while researching what different tech company's use I stumbled acrossthis article.
If it fails I think they should immediately put up donation funding on their website. The longer they wait between the failed KS and their website, the more people that might lose interest. Get it up and continue to improve it. The website will allow them to make changes to rewards unlike KS.
Damn, I miss BSG. See, if Brad had taken this idea and done that MMO he would have funded!
/thread hijack over
Pantheon: Rise of the fallen Kickstarter
If we learn anything from history we need to tell them now what we think would be good cause they wont ask for advice. with that being said heres what i came up with during my lunch break.
The first thing they need to do is not rush out the next form pledging. They need to learn and understand that if they take the time to get things right now it could get them millions more in pledges and crowd funding. They need to learn from other kickstarters and crowd funding games to see what has worked. In my option good models to look at are Star Citizen and Pathfinder. Star Citizen gave a good incentive to be an early backer and the ability to track and manage your pledge on their website. Pathfinder was very open with the financial side of their project and clear about what money would be going to and how it would be spent.
With that being said this is what I think the pantheon team should look to do. They need to make the barrier of entry to pledging as low as they can. This gets people financially invested into the game even if it is only a small amount. Once people are invested in a game they will invest more over time. Its just human nature. With that being said how can you do this? This is my proposal.
First things first Visionary Realms needs to work out a few things. They need to get the design Doc to the point that they can release Race/Class Combos. I have a made a quick example based off released information below.
WARNING GIANT WALL OF HARD TO READ TEXT
Race/Class combos.
Now they have this they need to get people to pledge. Now this requires a good website like star citizens. This might not be cheap but you need to spend money to make money! The website needs to have the ability to manage peoples pledges as well as "melt them" for store credit like Star citizen.
The first part of a pledge is the base pledge. This needs to be one Race and one Class. This could be sold as just a list or if they took the time to do it right some form simple character select screen on the website. This could start with just text and be worked on to include concept art, decryptions and eventually models.
Base Pledges: $20
ONE Race and ONE Class of your picking
(Access to tavern/housing module and other testing modules)
(First come first served invite to alpha and beta)
Picking your Pledge:
This needs to be as low as possible. The only limitation needs to be the amount of people they feel combatable having in an Alpha and Beta. This could also start at a lower price for early backers. But in my option keeping it as low as they can as it will get people to invest and hopefully hook them in to add more to their pledge at a later date. They might not like the game working this way but if you want a crowd funded game your going to have open it up for the world to see.
Next they need to look at the add-ons. This should be split into in game add-ons, Design add-ons and physical/digital add-ons.
In Game Add-ons
The races and Classes should all cost the same and be mange on the website. This is where the website needs to be clever and not allow people to buy races and classes they already own. I would keep the in game items to a minimal or not have them at all. They can always be added at a later date if needed. But best to keep the Pay-Win crowd to a minimal especially at the begging. I'm against the idea of a sub for life as I feel people willing to pay for it are the ones willing to sub to the game anyway and its only going to hurt your income in the future and pigeon hole you in future plans on finance.
Race add-ons: $10 each
Dwarf
Elf
Human
Ogre
Revenant
Class Add-ons: $10 each
Cleric
Crusader
Dark Knight
Enchanter
Rogue
Shaman
Wizard
Other Add-ons
Extra Character slot $5 each (Max 10)
*Housing fluff
Next is the Design add-ons, theses need to limited for obvious reasons and I don't see how it helps the game to have 1000's of player created items and this gives incentive for people to buy them before they run out.
Design Add-ons:
Name an item
Design an item
Name a NPC
Name point of interest
Design a quest
Design a Boss
Design a dungeon or raid
Now for the Physical/digital add-ons.
Physical/Digital add-ons:
Physical:
Collectors box
Cloth map
Signed poster
mouse map
Digital:
Map
Lore books
Now for how to release, make and test the game. They need to come out with a realistic and open stretch goal chart. They need to give people some form of access to the game as early as they can. The hanger model in star citizen was a great way for people to see their pledges and increase their attachment to the game. Again its about getting people invested to bring in more money in the long term.
Stretch Goal Chart:
On a side not I would try to make as much of this interactive as you can. Have player vote on concept art, to pick the next race that will done or added and so on. This keeps people coming to the website and maybe getting them to add and up their pledges.
*Tavern/House module:
If it were me, I would go with instanced housing. This lets make a place players can enter in game and the ability to sell and tone of fluff items for them to place in their houses. Again this creates more money for the game and also increases player's attachment to the characters and the game. If having non-instanced housing is very important to your design doc then you could go with having a tavern as just a place for player to walk around with their characters with out the option of buying fluff. I Always think instanced housing with guild housing being non-instanced is better.
*Dungeon module:
This will work by players creating groups within the tavern module and porting them to an instanced dungeon for testing. The game would work as a group lobby game for testing purposes only. This might not be the most ideal way for testing but it will gather hype and interest into the game.
*Crafting module:
A Crafting station added to the player's tavern/housing module and the ability to test certain aspects of the crafting system. They could also add crafting classes like the combat classes. This adds another thing for people to pay for and collect.
Arena Module:
A PvP arena built for players to duel and fight each other and test PvP
Now they have this they need to get people to pledge. Now this requires a good website like star citizens. This might not be cheap but you need to spend money to make money! The website needs to have the ability to manage peoples pledges as well as "melt them" for store credit like Star citizen.
The first part of a pledge is the base pledge. This needs to be one Race and one Class. This could be sold as just a list or if they took the time to do it right some form simple character select screen on the website. This could start with just text and be worked on to include concept art, decryptions and eventually models.
Base Pledges: $20
ONE Race and ONE Class of your picking
(Access to tavern/housing module and other testing modules)
(First come first served invite to alpha and beta)
Picking your Pledge:
This needs to be as low as possible. The only limitation needs to be the amount of people they feel combatable having in an Alpha and Beta. This could also start at a lower price for early backers. But in my option keeping it as low as they can as it will get people to invest and hopefully hook them in to add more to their pledge at a later date. They might not like the game working this way but if you want a crowd funded game your going to have open it up for the world to see.
Next they need to look at the add-ons. This should be split into in game add-ons, Design add-ons and physical/digital add-ons.
In Game Add-ons
The races and Classes should all cost the same and be mange on the website. This is where the website needs to be clever and not allow people to buy races and classes they already own. I would keep the in game items to a minimal or not have them at all. They can always be added at a later date if needed. But best to keep the Pay-Win crowd to a minimal especially at the begging. I'm against the idea of a sub for life as I feel people willing to pay for it are the ones willing to sub to the game anyway and its only going to hurt your income in the future and pigeon hole you in future plans on finance.
Race add-ons: $10 each
Dwarf
Elf
Human
Ogre
Revenant
Class Add-ons: $10 each
Cleric
Crusader
Dark Knight
Enchanter
Rogue
Shaman
Wizard
Other Add-ons
Extra Character slot $5 each (Max 10)
*Housing fluff
Next is the Design add-ons, theses need to limited for obvious reasons and I don't see how it helps the game to have 1000's of player created items and this gives incentive for people to buy them before they run out.
Design Add-ons:
Name an item
Design an item
Name a NPC
Name point of interest
Design a quest
Design a Boss
Design a dungeon or raid
Now for the Physical/digital add-ons.
Physical/Digital add-ons:
Physical:
Collectors box
Cloth map
Signed poster
mouse map
Digital:
Map
Lore books
Now for how to release, make and test the game. They need to come out with a realistic and open stretch goal chart. They need to give people some form of access to the game as early as they can. The hanger model in star citizen was a great way for people to see their pledges and increase their attachment to the game. Again its about getting people invested to bring in more money in the long term.
Stretch Goal Chart:
On a side not I would try to make as much of this interactive as you can. Have player vote on concept art, to pick the next race that will done or added and so on. This keeps people coming to the website and maybe getting them to add and up their pledges.
*Tavern/House module:
If it were me, I would go with instanced housing. This lets make a place players can enter in game and the ability to sell and tone of fluff items for them to place in their houses. Again this creates more money for the game and also increases player's attachment to the characters and the game. If having non-instanced housing is very important to your design doc then you could go with having a tavern as just a place for player to walk around with their characters with out the option of buying fluff. I Always think instanced housing with guild housing being non-instanced is better.
*Dungeon module:
This will work by players creating groups within the tavern module and porting them to an instanced dungeon for testing. The game would work as a group lobby game for testing purposes only. This might not be the most ideal way for testing but it will gather hype and interest into the game.
*Crafting module:
A Crafting station added to the player's tavern/housing module and the ability to test certain aspects of the crafting system. They could also add crafting classes like the combat classes. This adds another thing for people to pay for and collect.
Arena Module:
A PvP arena built for players to duel and fight each other and test PvP
Denaut
Trump's Staff
- 2,739
- 1,279
Hey you lied. You said that was hard to read!WARNING GIANT WALL OF HARD TO READ TEXT
Race/Class combos.
Now they have this they need to get people to pledge. Now this requires a good website like star citizens. This might not be cheap but you need to spend money to make money! The website needs to have the ability to manage peoples pledges as well as ?melt them? for store credit like Star citizen.
The first part of a pledge is the base pledge. This needs to be one Race and one Class. This could be sold as just a list or if they took the time to do it right some form simple character select screen on the website. This could start with just text and be worked on to include concept art, decryptions and eventually models.
-----------------------------------------------------------------------------------------------------------
Base Pledges:
ONE Race and ONE Class of your picking
Picking your Pledge:
This needs to be as low as possible. The only limitation needs to be the amount of people they feel combatable having in an Alpha and Beta. This could also start at a lower price for early backers. But in my option keeping it as low as they can as it will get people to invest and hopefully hook them in to add more to their pledge at a later date.
-----------------------------------------------------------------------------------------------------------
Next they need to look at the add-ons. This should be split into in game add-ons, Design add-ons and physical/digital add-ons.
-----------------------------------------------------------------------------------------------------------
In Game Add-ons
The races and Classes should all cost the same and be mange on the website. This is where the website needs to be clever and not allow people to buy races and classes they already own. I would keep the in game items to a minimal or not have them at all. They can always be added at a later date if needed. But best to keep the Pay-Win crowd to a minimal especially at the begging. I?m against the idea of a sub for life as I feel people willing to pay for it are the ones willing to sub to the game anyway and its only going to hurt your income in the future and pigeon hole you in future plans on fianc?.
Race add-ons:
Dwarf
Elf
Human
Ogre
Revenant
Class Add-ons:
Cleric
Crusader
Dark Knight
Enchanter
Rogue
Shaman
Wizard
Other Add-ons:
Extra Character slot
*Housing fluff
-----------------------------------------------------------------------------------------------------------
Next is the Design add-ons, theses need to limited for obvious reasons and I don?t see how it helps the game to have 1000?s of player created items and this gives incentive for people to buy them before they run out.
-----------------------------------------------------------------------------------------------------------
Design Add-ons:
Name an item
Design an item
Name a NPC
Name point of interest
Design a quest
Design a Boss
Design a dungeon or raid
-----------------------------------------------------------------------------------------------------------
Now for the Physical/digital add-ons.
-----------------------------------------------------------------------------------------------------------
Physical/Digital add-ons:
Physical:
Collectors box
Cloth map
Signed poster
mouse map
Digital:
Map
Lore books
-----------------------------------------------------------------------------------------------------------
Now for how to release, make and test the game. They need to come out with a realistic and open stretch goal chart. They need to give people some form of access to the game as early as they can. The hanger model in star citizen was a great way for people to see their pledges and increase their attachment to the game. Again its about getting people invested to bring in more money in the long term.
-----------------------------------------------------------------------------------------------------------
Stretch Goal Chart:
On a side not I would try to make as much of this interactive as you can. Have player vote on concept art, to pick the next race that will done or added and so on. This keeps people coming to the website and maybe getting them to add and up their pledges.
-----------------------------------------------------------------------------------------------------------
*Tavern/House module:
If it was me, I would go with instanced housing. This lets make a place players can enter in game and the ability to sell and tone of fluff items for them to place in their houses. Again this creates more money for the game and also increases player?s attachment to the characters and the game. If having non-instanced housing is very important to your design doc then you could go with having a tavern as just a place for player to walk around with their characters with out the option of buying fluff.
-----------------------------------------------------------------------------------------------------------
*Dungeon module:
This will work by player creating group within the tavern module and porting them to an instanced dungeon for testing. The game would work as a group lobby game for testing purposes only.
-----------------------------------------------------------------------------------------------------------
*Crafting module:
A Crafting station added to the players tavern/housing module and the ability to test certain aspects of the crafting system. They could also add crafting classes like the combat classes. This adds another thing for people to pay for and collect.
-----------------------------------------------------------------------------------------------------------
Arena Module:
A PvP arena built for player to duel and fight each other and test PvP
-----------------------------------------------------------------------------------------------------------
RDBMS is going the way of the dinosaur and so are traditional DBA jobs. Mainly because DBAs got paid an absolute fuckton just to try and eek out performance on something that you basically get for free on NoSQL. Facebook, Amazon, Google and Twitter are largely driven on NoSQL solutions because they're just flat out better. They scale better and can retrieve quickly. Facebook has ANSI SQL access to their NoSQL data store (I believe they don't use HBase, not sure what they use) using Presto. Apache is pushing a technology called Impala which gives similar access to HBase. It is a C++ thread that runs on your HBase machines. Flat out if you're going to stick your heels in and believe RDBMS is going to stay around and companies will always need DBAs it seems short sighted. RDBMS is good for certain solutions but for large-scale and data rich environments they fall very, very short. Since we moved off Oracle we've lost at least half our DBAs -- possibly even more. We've tried transitioning them to writing MapReduce jobs on our Hadoop/HBase cluster with varied success. But now that trying to squeeze every last ounce of performance out of an RDBMS doesn't happen (you just use a NoSQL approach) knowing every little configuartion and index rule isn't required which is basically what a DBA did. Every competent programmer can stand up a MySQL instance and do all the drop/create statements that DBAs generally were in charge of doing.
The place I work used to use an Oracle back bone and last year we ditched it for HBase. The new solution is offering so much more access and manipulation of the data I can't believe it was ever locked away behind a poorly performing RDBMS.
This post is also ignoring the rise of triple and quad store technologies which are also gaining traction (we're going to move our search engine to a triple store here soon) which also offer SQL-like access to the data. Ours uses SPARQL.
The place I work used to use an Oracle back bone and last year we ditched it for HBase. The new solution is offering so much more access and manipulation of the data I can't believe it was ever locked away behind a poorly performing RDBMS.
This post is also ignoring the rise of triple and quad store technologies which are also gaining traction (we're going to move our search engine to a triple store here soon) which also offer SQL-like access to the data. Ours uses SPARQL.
- 7,547
- 11,830
How bad does that Battlestar Galactica game have to be for me to have never heard of it? Anyone else heard of it, or played it?
I'll tell ya what.....this conversation that's been going on for the last few pages is 10x more interesting then the KS discussion. I have no fucking clue what everyone is talking about, but its fascinating to read.
Are you kidding me? Oracle is for the behemoth corporations who light their cigars with 100 bills. It's never going away.
There are still tons of companies that run COBOL processes. There are COBOL programmers I know personally that get paid 2x as much (yes, double) as any modern programmer.
All I do is pvp and I think trying to get any sort of pvp for launch is a horrible idea. As is crafting. It essentially needs to be an adventuring and monster slaying game at release. Otherwise all you will learn is what a shitty 100 million dollar mmog is made for only 5 million.
Nobody wants their stuff cut. Everyone wants it to be a fully immersive world. It is not possible. Build it on the shoulders of adventurers. If pvpers and crafters don't play at launch so what? It is supposed to be niched and targeted. Doing a couple of things extraordinarily well beats the crap out of adding one hundred tiny pieces of shit to pretend you have a living breathing world and end up trying to cater to everyone.
I realize a lot of people would not pay for or play that. That was never the plan though. So it is time to get real of what the core audience you are looking at. Right now it is as broad as any other mmog.
The ONLY way this project was and will ever succeed is if it grows over time. That means shooting for two hundred subs at launch and holding on to half of them is an absurdist pipe dream.
You target the game for people who essentially enjoy combat and monster encounters over anything else. That is the core market and essentially what your game is at launch. From there you build out the world to make it more encompassing. You have a huge crafting add on. A huge commerce add on. Additional classes and races.
They have to do things different. They can't just create a really limited budget photocopy of a full scale mmog and roll it out the sane way. That is guaranteed failure.
Right now there is no niche and no focus. Literally every mmog mechanic is on the table. The only things eliminated are mostly cosmetic that likely cost nothing at all to implement.
Listen I want to be a pvp only halfling bard but would easily give that up to get a game that was actually going to be focused on adventure group play. Give me more release dungeons over pvp and more classes.
To me what they should be focused on is so clear and easily defined and what should be kicked into the post launch future is equally clear.
Any feature that is discussed will have 25 people saying it is a must have and reminds them of EQ.
There were two aspects to original year 1 eq. Adventuring (fighting mobs ) and socializing via chat channels. That is all pantheon needs to launch.
I assume they will go with their own site funding. The upside to that is as soon as the money is pledged it can be put to use and keep the project going.
Without a short term infusion of cash I don't see how they progress any further.
Personally I would love for them to do a proof of concept / combat simulator demo kickstarter. Do it for thirty days and use it to raise $100K - $150k. Then that demo could be used to launch another large kick start or an ongoing website funding campaign. Not sure that will happen or even know if it is the best idea.
It seemed pretty clear at the beginning at the ks they needed money ASAP to keep going and/or add more people. Taking donations/pledges on their site gives them cash instantly.
Soygen
The Dirty Dozen For the Price of One
- 28,330
- 43,182
Or a better idea would be to make a game that more than 2600 people want to play.
Or make a game for far less than the average MMO costs, to appease those 2600 people.
