|
I think that if you had sufficient disk space for immediate needs, then this would be very doable - specifically for a text based game with a reasonable amount of players. As for a 5 year period, I think thats veering into the pointless in worrying about. Hard drive space is quickly becoming cheaper and cheaper for larger and larger amounts of space. Where a year ago I would have bought a 120 GB drive where I needed one, now I am finding myself buying 200 GB drives.
You could easily reduce the disk space needs by having a binary marshalling format for logged game data. And then theres actual compression on top of that. But I have what I consider to be a reasonable suspicion that the majority of the data would be socialisation by the players, unrelated to the playing back of events that happened in the past. Perhaps it might even pollute the playing back of events. Unless you had one of those weird roleplaying games where you could enforce players entering in character communications and actions in a specific way where out of character ones could be filtered out and discarded - but I was always an preferer of LP type muds.
As for determining how a bug was exploited, I think that this while a nice brute force approach, is somewhat futile. Better would be to have decent heuristics for viewing persisted state. If you can run code to find anomalous state for or action by a character, then you can either be alerted to it, or can log (perhaps only related) actions by that character and cascade the logging to characters related to the instigator.
However, I must disclaim this with the statement that it has been several years since I have run a mud with active players, let alone logged into a mud. Although I doubt things have changed significantly, I feel obliged to state this.
|