News: Means fighting AO Database limitations

[Chrisax]

This week, we went close to Apocalyse...

Friday with Means - May 8th, 2009
This was an "interesting" week. "Interesting" is going to have to be the word of choice as anything else would possibly being sensored by my own moderators.

The issue:

When AO was in development way back in 1998 (date and license number determined by a small piece of paper found at a previous employee's home office.)...the decision was made to go with C-Tree to store our in game resources...that Database you all happily rip every time we release an update to see the new candy. Metaing was kind enough to point out at the end of last week that this version of C-Tree had a size limit of 2 Gigs. Size limit really meaning "put more in it and it will do very, very bad things." Our current resource database is currently hovering at about 1.9 Gigs.

My options were to never add new graphical resources to AO...or Panic. I chose to Panic.

In 1998 2 Gigs was "Forever!" amounts of space...today this is obviously no longer the case. C-Tree has performed well for us over the last 9 years or so...so we have never had any reason to contact them about anything. I was pleased to find out that the company still existed (a good start!) and that they had more recent versions of the format that supported a larger capacity (large files...they tell me.) and better yet that the conversion process was straight-forward and reasonably painless. The fact that new pricing of C-tree now indicated that the new cost would be $5 per player was surprising.

When I recovered enough to stand I was happy to find that the C-Tree european manager was more than willing to be flexible with pricing to allow us to continue to use their excellent product. I do owe him a T-Shirt and a tour of the FC Oslo offices when he is in town....thank you Franco.

Thank-you Macrosun for working to reverse engineer the "lost" client build code to accomodate this update to the client RDB. Not fun..we know..and we all appreciate it. A /facepalm emote inspired by this event will be coming to the AO client in the near future.

This has resulted in *gasp* delays this week. We had been hoping to have the first views of instanced Pande, balancing, a new apartment and other 18.1 updates on TestLive this week but unfortunately this will now be coming next week instead.

The good news is that it seems to be all in viable shape and I am looking forward to seeing it on Test next week. It is likely I will need to "borrow" some guilds from the live dimensions to test it fully...any volunteer guilds would be appreciated. *hint hint*

Have a great weekend!

Means

[Chrisax]

OK, I realize the above is maybe not clear to everyone.

Let's explain a bit.

AO uses a database structure and engine (the way items data is stored and the thing that allow to manipulate it and retrieve it) that is called C-Tree.

C-Tree is a name created to show a difference with the system called B-Tree used in a lot of databases, and which is free. (If you are wondering, B-Tree is named after R. Bayer, who described it in 1970 and 1971, before micro computers.)

C-Tree is a totally proprietary system supplied and developed by a company called... FairCom (no joke).

Contrary to B-tree, C-tree is not free... not free at all. It's a commercial product working on a licenses scheme, and in my humble opinion, a pretty expensive scheme.

When AO was created, C-Tree was chosen for relatively unknown reasons today, but probably because it was easier to implement in the code of a game, and was able to find info "on the fly" faster and better than B-Tree for game purposes. Well... at least this was (probably) true in 1999.

In 1999, the C-Tree system supported only 2GB of data. This was gigantic if you consider that the best hard drives barely offered a few GB of storage at that time. And the first AO database was like 300MB (quickly up to 400 if I remember well). No one could imagine the game would do much more than double.

The fact is, the AO database not only contains the items you use (armor, weapons, etc... but nanos are a much more complicated thing than just a database entry), but the AO database also contains things you don't use or even see: mobs weapons, but also portals, and invisible items required to make the game work, etc. And that's a lot. So each time the world of AO expands, the database expands, and that would be true even if no players items were added!

And now... 1.9GB has been reached. In other words... no other big expansion would be possible with the 1999 C-Tree engine. In fact, even a small expansion wouldn't be possible.

And where it gets worse is that the new graphic engine will require new additions to the database (for the items that make the world "work")...

So... someone please press the Panic Button now!

That's what Means had to do... he explained the rest: incredible licenses cost at start (5 dollars per player, including froobs!) and a better deal (hopefully) in the end. Busy week...

So AO will get a new C-Tree database engine.

Now... to the question "why not using another system than C-Tree", the answer is... it is implemented and it works, and, at the moment a huge work is in progress for a new graphic engine, starting something new for the database would be suicidal, and, also, the 2GB limitation would hit even before the new engine, anyhow. In other words, swallow the medicine even if its taste is awful...

[Grind42]

Has FC ever heard of Project Managers.... take a project asses whats needed where your timeline is IE how long its going to take to implement & what problems might occur.

To me this should have been thought of before they even proposed putting the new engine in place & now its a case of hindsight. Or Means trying to look like a hero getting it up & running & explaining to everyone why it is allready so behind schedule.

They really should pull thier thumb out thier ass hole IMO. This is so typical of them, starting something without completely looking at the full picture.

TBH I dont care for this type of explanation, they promise a new engine, they better deliver or they will loose most of the loyal players they do have.


I could also be totally missing the point!!! I R Trox afterall...

P.S Glad its sorted even if he did have to use bribery & corruption

[Chrisax]

Yes... sounds like a big mess... and it is... but also some points should be known, not to find excuses (some things are hardly understandable), but to explain the situation, especially to make clear what the current AO team is dealing with.

- AO lifetime was planned (by project managers) to be 4 years. After 4 years, game over. We are now at 9 years, something we won't complain about, but the original project was never designed to last that long. So during the first 4 years and even after, some things (many things in fact) were implemented as "anyhow this is not intended to last or to support further developments" features. The current team has to deal with this situation.

- After 4 years, it took a while to decide AO would be a long-lasting game, hence a lack of long-term planning.

- AO original development was like 45 coders and 20+ various designers, etc. The only one left (to my knowledge) is Enno Rehling but after a 3 years or so break from AO and FC, and, originally, he never worked, for example, on graphics programming, special effects, artificial intelligence, etc. And the teams he was in employed more than 25 persons so he simply never worked personally on many parts of the code.

- A part of the documented code was... lost with the devs. (Yes, this is bad and hardly understandable.) So it has now to be reverse-engineered. It's FC's problem of course, not players' problem. But the current team can't be blamed for that. They are in a mess.

As for Means' post, to the contrary of several former game directors, Means tells what is happening as he feels it and without any diplomatic language. He speaks of errors and issues too. He didn't have to tell us what he told us today. So, even if we can criticize the situation itself (and there are things to question, indeed!), I'm not sure we can blame him for being honest.

[Elrojo]

tbh it doesnt seem like a good plan, when you plan to fail. I dont understand why you plan to stop after 4 yrs? If you get to 3 yrs and the player number are very high then you either give them 1 yr to pack their bags or you re-evaluate and elongate the original plan. I dont understand why AFTER the cut-off time they decide "lets keep going". And then FIVE years after that go "uh oh". It does seem a little silly.

I guess to back up Grind its good that Means is telling us about the problems but in the end they are very avoidable problems. So just because he is telling us these things shouldn't then preclude us from being annoyed anyway.

To back up Chris, I have taken over operations at a pharmacy here that has long been poorly operated. I do get annoyed customers in who have been inconvenienced as a consequence of the OLD management and I get shot because i'm the messenger. It is hard to pick up a project in the middle and run with it as if you have been there from day 1.

I just hope this "little" problem triggers a very detailed look at ALL other aspects of the game/business/code etc and all other potential problems are investigated.

I propose however to delete Grind's bank which surely takes up half of the database

[Mophro]

Has FC ever heard of Project Managers.... take a project asses whats needed where your timeline is IE how long its going to take to implement & what problems might occur.

Yea like Project Managers solves all problems . In my reallife experience it could quite be the reverse .

[Elrojo]

dont they say a Camel is what a horse would look like if it were designed by committee (insert project managers)

[Rulzern]

tbh it doesnt seem like a good plan, when you plan to fail. I dont understand why you plan to stop after 4 yrs? If you get to 3 yrs and the player number are very high then you either give them 1 yr to pack their bags or you re-evaluate and elongate the original plan. I dont understand why AFTER the cut-off time they decide "lets keep going". And then FIVE years after that go "uh oh". It does seem a little silly.

This is generally how it's usually done in the MMO business, 4 years is a looooong time in the gaming world, and you can't really plan accurately for that long. You _can_ conceivably plan for an MMO to last 10+ years, but in the days that AO was made, Ultima Online had been running for one year, and Everquest came out the same year, there was simply no reason to beleive that AO would be sustainable for as long as it has been, and you don't build a castle when you need a bicycle shed.

Also, what were the alternatives in 1999? Did the alternatives suit the needs of a game like AO? I find it hard to beleive that they did what they did for no reason.

IMHO there are too many variables in the picture to say "they should have done X instead!", that AO is still alive after more than twice of the projected lifetime has passed, gives me the impression that they did more things right than most of the competition.

[Grind42]

IMHO there are too many variables in the picture to say "they should have done X instead!", that AO is still alive after more than twice of the projected lifetime has passed, gives me the impression that they did more things right than most of the competition.

I put this down to the hectic learning curve of the game & the community, its not a game you can jump in & become Uber within a month it takes time you also cant learn it in a month. I have been playing for 6yrs & I still learn things about the game on a month to month basis.

[Ceokittymeow]

IMHO there are too many variables in the picture to say "they should have done X instead!", that AO is still alive after more than twice of the projected lifetime has passed, gives me the impression that they did more things right than most of the competition.

I have to agree with Rulzie here as even though Ao was planned for a 4 year stint, it has lasted double the length and then some!

There are many games that have been published that have been and gone (Tabula Rasa to name one) that were decent games that had there stretch and have keeled over. For a online game that is being developed there are always timetables that they run as to make sure that they do not lose money on server operation, patches etc. on a product that wont rake in the funds.(though i still have no idea why games like Star Wars Galaxies is still running with the craptastic population there)

As for the C-tree database, I can see the major oversight about the size limitation on FC's AO database, and glad to see that they are working on getting a new RDB up and running to make sure that AO can expand w/o that problem at least... Though i question FC's awareness about catching it now and not before hand... Age Of Conan's database is a prime example... I have seen AO data in a AoC rip before... Now this is from here-say and from a pic i have seen from a firend who played it, but if FC did use part of AO database, wouldn't they have used the same style of database storage? And if so HOW THE HELL did they miss that?

Anyways i digress... I am glad to see that Chris, oops i mean Means is admitting that FC just avoided a sh*t storm and was nice enough to share that information with the community. I think that Means doing so is showing trust and loyalty to FC's AO Patrons and should keep us informed, though us patrons are twiddling down and out.

Dancing for FX upgrade pl0x! *keeps an eye on Chris's crafty subterfuge*

[Ceokittymeow]

DOUBLE POST!!!

I forgot to add that I am glad to see that 5 USD are going to fund the resource database that i use. It will make my time looking at Auno and the other DB sites more heart warming and fuzzy feeling that I am a part of keeping that DB alive