Go-ipfs: Performace, or How IPFS will be better than BitTorrent

[DonaldTsang]

Here are a list of issues that is plaguing IPFS: Performace. It has to beat existing system to gain a wider adoption. One factor is "the default should be semi-optimal", where once installed, without much parameter tweaking, it will perform better than BitTorrent.

• https://github.com/ipfs/go-ipfs/issues/2111 (transfer speed)
• https://github.com/ipfs/go-ipfs/issues/2828 (DHT management)
• https://github.com/ipfs/go-ipfs/issues/2917 (network traffic usage)
• https://github.com/ipfs/go-ipfs/issues/3429 (another bandwidth usage)
• https://github.com/ipfs/go-ipfs/issues/3545 (file adding)
• https://github.com/ipfs/go-ipfs/issues/3638 (import issue)
• https://github.com/ipfs/go-ipfs/issues/3885 (hanging system)
• https://github.com/ipfs/go-ipfs/issues/5221 (slow pinning)
• https://github.com/ipfs/go-ipfs/issues/5438 (another)
• https://github.com/ipfs/go-ipfs/issues/5723 (Bitswap)

Here are some tests

• https://github.com/ipfs/go-ipfs/issues/3131 (speed testing)
• https://github.com/ipfs/go-ipfs/issues/3316 (adding files)
• https://github.com/ipfs/go-ipfs/issues/3317 (high-speed ethernet)
• https://github.com/ipfs/go-ipfs/issues/3860 (IPNS)
• https://github.com/ipfs/go-ipfs/issues/5037 (another speed test)
• https://github.com/ipfs/go-ipfs/issues/5226 (newer test)
• https://github.com/ipfs/go-ipfs/issues/5420 (further analysis)

[Stebalien]

Yes we have performance issues and we're working on them. But this issue doesn't really add anything to the discussion.

[DonaldTsang]

@Stebalien I was expecting this to be a meta-issue on performance

[Stebalien]

Ah, got it. My concern with this issue is that it's kind of a grab bag of different unrelated performance issues and we already have a perf label.

There appear to be four categories:

• IPNS is slow: https://github.com/ipfs/go-ipfs/issues/3860
• Adding files to go-ipfs is slow. Could probably use a good meta issue, want to make one?
• Bitswap transfer speed is slow: https://github.com/ipfs/go-ipfs/issues/5723
• IPFS uses a lot of background bandwidth: https://github.com/ipfs/go-ipfs/issues/3429 (may need a better meta issue?).

[DonaldTsang]

@Stebalien In that case making a meta-issue for all four topics would be better, it would make reading easier.

[jasonzhouu]

@DonaldTsang
suggest a new item:

• https://github.com/ipfs/go-ipfs/issues/6382 (latency of DHT resolving)

change a item:

• https://github.com/ipfs/go-ipfs/issues/5226 (delay test)

[DonaldTsang]

@jasonzhouu are you suggesting to combine those two issues?

[jasonzhouu]

@DonaldTsang
I'm suggesting to:

• add the first one
• and change the description of the second one, which already exist in the original list.

[Stebalien]

I've created a meta issue for resolution delay #6383.

[jasonzhouu]

I've created a meta issue for resolution delay #6383.

Awesome!