How long should FAF keep old replays?
-
AI games can be things like team survival games. Some of those games are posted to YouTube. People care enough to put the videos online, that means people care about the games. Maybe very few people, other than the players themselves, care about solo games involving AI. But you can't say as a general rule that nobody cares about AI games.
The cost of making a backup of the games before deleting them from the server is minimal. What even is the cost of 300GB of cloud storage? Just save them somewhere. Maybe in 5 years the cost of storage will be so small that we can restore all the replays to the server at basically no cost. Or some other way to make the replays available. There is no good reason to delete history even if most of the history is crap. Once it is deleted we can never get it back.
Do we even have a complete backup of all replay files? Just load them into a single 1TB drive and send it to a FAFer's house. 1-2 copies like this would ensure that history is not lost.
-
Cloud storage for our purposes is around 5€ per month per 100GB. Previous calculation said we have a growth of about 20GB per month, so the costs are constantly increasing. Just keeping everything forever is not a viable option.
Also running it on a private copy is essentially the same as deleting for 99% of the users who just see "ah, replay not available, well bad luck".
-
Would it be possible to look up the storage savings from the various proposed solutions?
-
@Brutus5000 I wouldn't mind volunteering to be personally responsible for responding to people who want really old replay files for some reason. We wouldn't even need cloud storage, I would just keep a single 1TB or 2TB drive with zero ongoing costs to FAF. If the files were transmitted to me electronically it would be close to zero cost to FAF. That situation could then continue essentially indefinitely.
I doubt there is very much interest by players in those old replays, so we're probably talking about 1 request per month. Even if I have to spend 15 minutes a day responding to those old requests, I wouldn't mind.
-
Arma for The Giver 2021
-
arma for Archive Councilor 2021
-
Before deleting any replays, consider the following:
- New format for .fafreplay, compressed achive of the scfareplay-file and the json-data without the final b64-encoding
Comparsion of size: scfafreplay: 1222 KB fafreplay-current: 260 KB (21.2% of original size) fafreplay-zip: 196 KB (16.0% of original size) fafreplay-7z:: 141 KB (11.5% of original size)
-
Publish the replay vault (maybe IPFS?) so people can mirror / clone the replays for historic reasons.
-
Create a service that collects local replays from players to be able to repair current broken / missing ones in vault
EDIT:
Did some more testing and the compression numbers were a bit messed up earlier. I've updated the post now with my results. -
+1 on the service,
you could use some sort of cheaper object storage like AWS glacier or azure archive as place to dump it + a cron job to just upload xGB every + some sort of proxy api to pull the data when requested (probably almost never).EDIT: looks like wasabi could be relatively cheap https://wasabi.com/
-
I'm in favor for deleting them after 3 - 5 years. i don't value old replays as much because they were played with an older version of the game. Very good replays will be casted most likely. Also if you want to watch old gpgnet replays https://www.gamereplays.org/supremecommanderforgedalliance/replays.php?game=44
-
Hi guys, thanks for your feedback so far.
Fortunately we finalized some work on replay stuff that was mostly done already, so we'll compress replays with zstandard instead of (zip encoded with base64). Support will come in the next client release and we'll need to find a way to convert all the old replays. This will be a big breaking change for some tools out there, but it's the best chance we have.
Some words on the alternative suggested here:
While renting storage in the cloud with some cheap providers is possible of course it massively increases develop and (more important) maintenance efforts which I'm not willing to pull. Storage in our Hetzner datacenter would be possible though as it can be natively integrated into the existing server infrastructure. Nevertheless we still strive for a cost neutral solution if possible.Eventually this topic will reappear in a few years again, so the discussion is not off the table forever.
We'll see which storage savings we can achieve with the current solution. -
Will this break the replay parser?
-
@FemtoZetta said in How long should FAF keep old replays?:
Will this break the replay parser?
Perhaps, but it should not be difficult to find a Java library that can decompress the ztd file format and add that to the parser itself. So it shouldn't be difficult to update the parser to handle this. If anyone is actively maintaining the parser, it should be an easy fix.
Even if there's no way to update an existing tool, it should be relatively easy to decompress the ztd files "by hand" and re-compress them in the fafreplay file format in order to hand them off to a software tool.
OR someone could even make a software tool to convert replay files in the new (ZTD) format into the old (zip) format. Automating the process of converting files from a more-efficient format to a less-efficient format so that old tools can use them is kind of silly but it would solve the immediate problem of tools being broken.
-
What is "the replay parser"?
-
-
Yes, it will break for the new format. Someone will have to update it, somewhere around line 430 in page source. This diff should help with that.
-
Oh please keep all replays. Can we throw money at buying more storage?
-
Imo, no need to keep replays older than 3 years no matter what they are. If they were good they are already casted w/e.
Even then I wouldn't really keep replays older than 2 years unless they are from ranked games. It just seems pointless to keep so much when it's useless. -
I would suggest to keep only the special old games. But I think there are no stamps like "Tournament". I also think that no one will sort their favorite replays. I guess you can just save replays of certain players, purely for memory?
But in general, I do not see any point in these replays, the game is growing up, the old tactics and methods of the game can not be applied to the current realities.
I am in favor of removing all replays since the last old patch.
And further sorting them with some parameters
- Removing replays with one player
- Removing replays with a sandbox
- Removing of replays of the company
- Removing "blacklisted" maps (gap, astro, etc)
- Removing replays with desync
- Removing replays after (optional) 4/8/12 months
-
@ZLO I've been searching for this parser for ages! The old link on google 404s
edit: still seems to be broken for me -
@Cascade You should try to open it with vpn. Works fine for me without vpn