Slow downloads of TrueNAS updates (Switzerland)

Yeah, so basically it is just another unprofitable S3 customer.

Seriously? That would explain why their iso download is so bad :laughing:
I don’t remember them having such a bad performance in the past!

Also not a very smart business decision in my opinion, there are faster and cheaper options like Cloudflare.

Do you have problems with downloads? As far as I remember, you are from Switzerland, what’s your ISP and could you please provide MTR to the download.truenas.com?

For most locations the speed is faster, but sometimes the routes could be wrong.

Correct, Init7.

  7     8 ms     8 ms     8 ms  gw-corebackbone.init7.net [77.109.134.191]
  8    13 ms    13 ms    13 ms  ae12-2022.fra10.core-backbone.com [81.95.15.57]
  9    13 ms    13 ms    13 ms  coreb-fra.cdn77.com [169.150.195.68]
 10    13 ms    13 ms    14 ms  vl217.fra-cyx2-dist-2.cdn77.com [169.150.194.51]
 11    13 ms    13 ms    13 ms  136.0.77.2

No working IPv6, why am I not surprised?

Because most of European traffic relays cannot agree with each other and we have had problems with a IPv6 traffic in Europe. So right now it’s better to have it disabled at least for Europe.
I passed your info to the team. Doesn’t look like there should be any issue.
What’s your internet speed and what’s speed when you download a TrueNAS update?

In Italy, I can download easily from TrueNAS buckets (Storj obviously) at 400-500 mbps.

2 Likes

Today it is difficult to make a screen, it fluctuates between 20 and 40 Megabytes/s. However, it takes less than a minute.

3 Likes

20MB/s on a 10Gbit/s line.

20 to 40 Mb/s is not 400-500Mbit/s. That is 160 to 320Mbit/s.

1 Like

Of course!
Still the network in Italy is a bit congested, due to the Easter holidays.

A couple of months I urgently needed the iso, and downloaded it to a usb 3 stick with ventoy. Up to 60 MB/s.

I share large files with storj, and everyone is impressed with the speed.

4 Likes

Isn’t that what “decentralization” promised to solve?

I just think it is kind of funny that we went from
“faster than traditional”
to
“not actually decentralized, just an S3 with nodes as storage backbone”
to
“we actually can compete with traditional S3 performance, but everyone is super impressed with the speed”

Sometimes, it depends on ISPs and their routing. Even Steam, AWS, Google Cloud are strongly influenced. P2P is even more penalised due to traffic shaping by ISPs.

2 Likes

True. It also depends on the ISP peering. The thing is, Init7 is the greatest ISP there is, with the greatest peering there is. So if your service is well peered to DE-CIX, it will have great performance on Init7, because they have great peering to DE-CIX.

No. Great ISPs like Init7 have steam, and netflix and linux mirros and many other cache server in their network. If you don’t get 1Gbit on steam, your ISP sucks. I get around 9Gbit.

Ohh boy this is so confidently wrong, I don’t even know where to beginn.

  • P2P in the STORJ context should be faster, because you are less depended on good peering. I don’t need good peering to from Init7 to DE-CIX to T-Mobile US, because thanks to decentralization, I could get best case from another Init7 node or second best case a node with great peering to Init7 (or basically to DE-CIX) like Telekom Germany.
  • I doubt there is a lot of traffic shaping going on, we don’t have 2005. Maybe some shitty US ISPs slow down some ports. Which brings me to the next point
  • ISPs can’t traffic shape my TrueNAS download, because that happens over port 443.

TLDR: In my opinion STORJ in Europe has poor performance and is overpriced. I hope they soon can no longer afford to offer S3, that would kill one distraction and bring us closer to a subsidies free future where we will finally see if this works.

1 Like
## Zurich 2 136.0.77.2 [EGIHosting (USA)]

# Mtr Start: Thu Apr  4 14:22:27 2024
HOST: 10g-zur-kvm-1-89-187-165-1. Loss%   Snt   Last   Avg  Best  Wrst StDev
  1. AS60068  89.187.165.28        0.0%    10    0.2   0.1   0.1   0.3   0.0
  2. AS60068  138.199.0.178        0.0%    10    0.3   0.2   0.1   0.3   0.0
  3. AS174    149.11.89.129        0.0%    10    0.4   0.5   0.3   0.8   0.0
  4. AS174    130.117.49.165       0.0%    10    1.1   1.0   0.9   1.5   0.0
  5. AS174    130.117.0.18         0.0%    10   13.3  10.6   5.8  24.3   6.6
  6. AS174    154.54.36.53         0.0%    10   11.8  11.6  11.3  12.3   0.0
  7. AS174    149.11.183.154       0.0%    10   10.4  10.5  10.4  10.7   0.0
  8. AS60068  169.150.194.53       0.0%    10   11.0  10.9  10.8  11.0   0.0
  9. AS18779  136.0.77.2           0.0%    10   10.9  10.9  10.8  11.0   0.0

# PING 136.0.77.2 (136.0.77.2) 56(84) bytes of data.
64 bytes from 136.0.77.2: icmp_seq=1 ttl=60 time=10.8 ms
64 bytes from 136.0.77.2: icmp_seq=2 ttl=60 time=10.9 ms
64 bytes from 136.0.77.2: icmp_seq=3 ttl=60 time=10.9 ms
64 bytes from 136.0.77.2: icmp_seq=4 ttl=60 time=10.9 ms
64 bytes from 136.0.77.2: icmp_seq=5 ttl=60 time=10.9 ms
--- 136.0.77.2 ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4003ms
rtt min/avg/max/mdev = 10.813/10.873/10.922/0.037 ms

# traceroute to 136.0.77.2 (136.0.77.2), 30 hops max, 60 byte packets
 1  unn-89-187-165-28.cdn77.com (89.187.165.28)  0.104 ms unn-89-187-165-29.cdn77.com (89.187.165.29)  0.104 ms unn-89-187-165-28.cdn77.com (89.187.165.28)  0.117 ms
 2  vl201.zur-itx1-core-1.cdn77.com (138.199.0.178)  0.123 ms  0.126 ms  0.136 ms
 3  hu0-1-0-2.rcr51.b021037-0.zrh02.atlas.cogentco.com (149.11.89.129)  0.403 ms  0.432 ms  0.452 ms
 4  be2395.ccr52.zrh02.atlas.cogentco.com (130.117.50.25)  1.174 ms  1.163 ms be2387.ccr51.zrh02.atlas.cogentco.com (130.117.49.165)  0.962 ms
 5  be3073.ccr22.muc03.atlas.cogentco.com (130.117.0.62)  6.012 ms  6.075 ms be3072.ccr21.muc03.atlas.cogentco.com (130.117.0.18)  5.976 ms
 6  be2959.ccr41.fra03.atlas.cogentco.com (154.54.36.53)  11.519 ms be2960.ccr42.fra03.atlas.cogentco.com (154.54.36.253)  11.759 ms  11.518 ms
 7  149.11.183.154 (149.11.183.154)  10.533 ms  10.568 ms  10.457 ms
 8  vl215.fra-cyx2-dist-1.cdn77.com (169.150.194.49)  10.809 ms vl217.fra-cyx2-dist-2.cdn77.com (169.150.194.51)  10.916 ms vl215.fra-cyx2-dist-1.cdn77.com (169.150.194.49)  10.918 ms

Perhaps TCP congestion. :- )

I know it sounds absurd, but some providers in Italy, especially Vodafone Italy, nail down p2p. In the past, Tele2 (later acquired by VF), blocked connections completely and for example Minecraft wouldn’t even work!

It is now very rare, but with some small European providers even downloading the iso of ubuntu is painful. I hope they stop in the name of net neutrality! Sometimes I share LEGAL files via p2p at ridiculous speeds.

I am closing this discussion as it is a borderline case (users complaining about the ridiculous speeds of storj). I am a strong supporter of decentralisation! In addition to being a SNO, I am an end user :slight_smile:

P.S. I have an ‘Aruba Broadband’ connection, literally provided by the well-known Italian hosting company. Literally allows its customers to have servers at home.

1 Like

Sure! But then again, it is not possible by slowing down port 443.

Again, if your ISP is so shitty it does not provide its own mirror, yes that is true. Good ISPs on the other hand have an ubuntu mirror. I download every major Linux ISO and update at 10Gbit/s. I mean, Init7 has what?Maybe 100k customers? It is not that hard!

I also don’t care about problems that don’t affect me /s

So you dislike the centralized TrueNAS S3 way and support an iso torrent instead? Good! Me too :smile:

This is relevant how? Sorry I don’t understand the point you are trying to make.

1 Like

Aruba is not primarily an ISP, but a hosting company. It has fabulous routing to host a server at home. I am closing this discourse here because every country has its own providers :slight_smile:

Otherwise I support your argument! I don’t understand why some people complain about dial-up speeds on s3 xD

1 Like

I always thought that datapacket is the best service out there if anyone is looking for low-letancy and consistent delivery. It is the first time I hear about Aruba and Init7. Literally it seems to look that Zurich is not so well connected. From Frankfurt to EgiHosting on DataPacket I am getting 0.5 ms pings which is more then half of my ping from a leading large cloud provider with a datacenter directly located in Frankfurt and about 20-30 times better what I noticed is possible from Zurich. I am smiling a little bit of course. :- )

Well, if they experience dial-up speeds on S3, I totally understand why the complain about dial-up speeds on S3.

No. Zurich is just some of their knots. Also Zurich is not important.
What is important if you are doing business in Europe, is the by far biggest interchange DE-CIX in Frankfurt. So my ISP has great peering to DE-CIX. That means that STORJ has bad peering to DE-CIX. Simple as that.
Here is another example to that explains it. Imagine we do a video call. I am on Init7 and you are on Telekom DE. We probably don’t have a direct peering to each other. So we go through DE-CIX.

That is why all major Europan services have a good connection to DE-CIX.
For example Hetzner, has 2800Gbit/s peering to DE-CIX.

STORJ needs good peering to DE-CIX. Just like a Swiss IPTV provider Zattoo needs good peering to the Swiss equivalent, so all Swiss customers can experience great streaming. Or you go the extra mile and even offer direct peering. Which of course would be even better but not realistic for STORJ.

1 Like

So to get a little bit back on topic, are these bad speeds something the team is looking into?
Or do you guys think, this is good enough?

Ah, now I think I understand what you had on your mind.

Where is storj “hosted” currently? What do you think of moving storj main servers to DataPacket? Would the peering improve? But look, to be honest I think it does not matter that much (I mean it does but not that much) because it seems that the main bottleneck is not that much the location of storj central servers but the structure of the underlining network. Take a look for example on the charts below that are based on the payment distribution for the month of January 2024 with STORJ/USD exchange rate as of a date of payment announcement (beginning of February, dont remember the exact date). :- )



You may also take a look at @Ambifacient charts as those are his stat arbitrage skills (mostly Extended Kalman Filter and Vine Copula Mispricing Index Strategies) to be blamed for all this mess. :- )


And also compare with the current storj map as presented on their network stats map here. :- )