Can't Achieve 'Great' Score for Video Streaming on Cloudflare Test

My Cloudflare speed test shows "Great" for gaming and video chat, but only "Good" for video streaming, even with SQM enabled. Any suggestions to optimize for streaming?

$ cat /etc/config/sqm

config queue 'eth1'
        option enabled '1'
        option interface 'pppoe-wan'
        option download '95000'
        option upload '50000'
        option qdisc 'fq_codel'
        option script 'simplest_tbf.qos'
        option linklayer 'ethernet'
        option debug_logging '0'
        option verbosity '5'
        option overhead '44'
        option linklayer_advanced '1'
        option tcMTU '2047'
        option tcTSIZE '128'
        option tcMPU '84'
        option linklayer_adaptation_mechanism 'default'

Since cloudflare's scoring algorithm has, to my knowledge, not been revealed I think we could at best speculate wildly...

Correction:
https://blog.cloudflare.com/aim-database-for-internet-quality

Seems to contain relevant information, but too late for me today...

1 Like

You need ~50Mbps for 8k streaming.

It would be 'great' if you had more bandwidth.

I find that it is best to evaluate based on real world experiences, rather than some artificial “score”. Are you actually having steaming video performance issues?

1 Like

Strangely, I get a "Great" score on all three metrics when SQM is disabled. :upside_down_face:

Sometimes, my YouTube video streams default to 720p. I'm not sure why, but it could be due to poor routing by my ISP.

That can happen for a lot of reasons - including congestion relief from the YouTube servers (not related to your local network connection). But as long as when you set it to 1080p it works well, you’re not likely having any real network issues.

1 Like

You doubled your throughput. Now you have more than enough overhead to get 'great'.

Unfortunately, the CPU (MT7621AT) couldn't handle more than that.

Can you backtrack and illustrate the problem that you solved using sqm?

Metric 0 points 5 points 10 points 20 points 30 points 50 points
Loss Rate > 5% < 5% < 1%
Jitter > 20 ms < 20ms < 10ms
Unloaded latency > 100ms < 50ms < 20ms < 10ms
Download Throughput < 1Mbps < 10Mbps < 50Mbps < 100Mbps < 1000Mbps
Upload Throughput < 1Mbps < 10Mbps < 50Mbps < 100Mbps < 1000Mbps
Difference between loaded and unloaded latency > 50ms < 50ms < 20ms < 10ms

Formula for streaming: download bandwidth + unloaded latency + packet loss + (loaded latency - unloaded latency difference

Based on the formula, my guess is I need a download speed exceeding 100Mbps to achieve close to 70 points for a 'Great' result.

1 Like

So according to this you get scores of:

  • Streaming: download bandwidth + unloaded latency + packet loss + (loaded latency - unloaded latency difference):
    The score to qualifier mapping is:
    bad: 0-10
    poor: 10-15
    average: 15-35
    good: 35-60
    great: 60+

20 + 5 + 10 + 20 = 55 -> you seem to be ~5 points short of being classified 'great', either increasing the download capacity above 100mbps or reducing the unloaded latency to < 20ms would do the trick

  • Gaming: packet loss + unloaded latency + (loaded latency - unloaded latency difference):
    The score to qualifier mapping is:
    bad: 0-5
    poor: 5-15
    average: 15-20
    good: 20-30
    great: 30+

10 + 5 + 20 = 35 -> 'great'

  • RTC/video: packet loss + jitter + unloaded latency + (loaded latency - unloaded latency difference):
    The score to qualifier mapping is:
    bad: 0-5
    poor: 5-10
    average: 10-20
    good: 20-30
    great: 30+

10 + 10 + 5 + 20 = 45 -> 'great'

This explains the score... BUT I would not put too much stock into these classifiers, sure conceptually this makes some sense as few end users intuitively can interpret the raw numbers well, but whether you get a great or a a good really makes little difference IMHO.

As @frollic implied, as long as you perceive your network to be fine no reason to stress about cloudflare's score here. Keep in mind, that while it is great that cloudflare offers this service (I especially applaud them for measuring and reporting working latency), cloudflare has its own perspective/agenda here (they are after all in the business of convincing ISPs to aim for high capacity low latency interconnects with their own CDNs). IMHO that (i.e. having their own agenda) is A-OK, just something to keep in mind when interpreting their scores.

2 Likes

It all makes sense. Thanks for explaining.

This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.