Web Squad ISP

[Mr Scratch]

What speed do you get off of this? This is Apple IPSW locally cached. This should indicate your max attainable line speed on local.

http://updates-http.cdn-apple.com/2...412B0A59/iPhone11,8_12.1.4_16D57_Restore.ipsw

 

[Mr Scratch]

Thanks, wasn't the answer I was looking for, more like one aimed at the layman.

Get any proper, 1Gbps router. You're trying to use a router designed for LTE on 1Gbit fibre. And don't expect to get good performance over WiFi. Manage your expectations, please.

 

[wingnut771]

Sorry I haven’t been keeping track, are you not happy with your 1gbps package? I’m considering moving from CISP to websquad 1gbps but now not so sure.

No, my router (looking for a new one) and now last mile is the problem.

 

[wingnut771]

Get any proper, 1Gbps router. You're trying to use a router designed for LTE on 1Gbit fibre. And don't expect to get good performance over WiFi. Manage your expectations, please.

 

[Mr Scratch]

and can you drop the hostility a bit.

When you're flooding the thread with arbitrary tests, I will do and say as I so please.

 

[wingnut771]

When you're flooding the thread with arbitrary tests, I will do and say as I so please.

Ok, I'll leave you alone.

 

[Concentric]

Sounds like it's not automatically negotiating 1 Gbps. Any way you can force 1 Gbps within the Ethernet settings on the Huawei?

I've checked from our end, we can't provision the other ports. Can you send a ticket to support so we can track this?

also how cant this be the router. when plugging pc directly to vuma box on ge1 100mbps on ge2 1gbps link.

 

[Mr Scratch]

also how cant this be the router. when plugging pc directly to vuma box on ge1 100mbps on ge2 1gbps link.

Is this a new issue? The ONT could be improperly negotiating the rate on that one port, which requires Vuma to reconfigure/reset that port.

 

[Concentric]

Is this a new issue? The ONT could be improperly negotiating the rate on that one port, which requires Vuma to reconfigure/reset that port.

same issue as before, just narrowed it down to ont, not router.

 

[sand_man]

I'm going to stop now and wait for local iperf.

Cat6 cable? You got another you can test? Performance looks way off dude...

 

[websquadza]

also how cant this be the router. when plugging pc directly to vuma box on ge1 100mbps on ge2 1gbps link.

The ZNT ONT may be offering FE (100mbps) instead of GE (1Gbps) for some unknown reason- this we’ll ask Vumatel to check and maybe re-provision. The Huawei device might have a GE port, but it’s probably set to Auto Detect, forgoing any negotiations. Some routers let you specify 1G and the other device is forced to listen. Anyhow, this will need to be taken up with Vuma.

Also follow up with the team regarding your 1G router options.

Thanks

 

[Concentric]

The ZNT ONT may be offering FE (100mbps) instead of GE (1Gbps) for some unknown reason- this we’ll ask Vumatel to check and maybe re-provision. The Huawei device might have a GE port, but it’s probably set to Auto Detect, forgoing any negotiations. Some routers let you specify 1G and the other device is forced to listen. Anyhow, this will need to be taken up with Vuma.

Also follow up with the team regarding your 1G router options.

Thanks

Thanks! will do

 

[Concentric]

@websquadza


rebooted ont.
Did the trick
As a side point would google drive be considered local traffic?
only getting around 20Mbps.

Anyways, yay awesome.
And thanks so much for your weekend response.
One of the many reasons i picked WebSquad

 

[Mr Scratch]

@websquadza


rebooted ont.
Did the trick
As a side point would google drive be considered local traffic?
only getting around 20Mbps.

Anyways, yay awesome.
And thanks so much for your weekend response.
One of the many reasons i picked WebSquad

Testing Google Drive performance

Preface
Recently I've been getting into storing my backups offsite since I now have the bandwidth to do so. I chose Google Drive as I could get unlimited storage via their GSuite offering, which I use for personal mail. For this reason, it should be expected that I require as much bandwidth as possible to complete these backups in a timely manner.


Initial testing
I'll be conducting tests using a linux based server directly connected to the router via cable, the gdrive-cli client (here: https://github.com/prasmussen/gdrive) and my 200Mbit Cool Ideas line.

I started by first creating a 1Gb sparse file which I would use to benchmark the speeds, this will eliminate factors such as file system speed and disk fragmentation.

View attachment 491247

I then ran the client using the pure stock configuration, and logged the results via my firewall (to compare to the results reported via gdrive-cli)

View attachment 491249

Hmmm, not quite as I expected. I fired up Wireshark and looked for the server I was uploading to and found this: 216.58.223.42

View attachment 491251


Well then, the server is local which rules out any TCP window problems. I spun up DSTAT for an overview of the system's performance during the test just to see that I was not maxing out any other piece of hardware and thats when I noticed something interesting. My box wasn't sending data continuously but rather send out chunks at a time, and the gdrive-cli client simply gave me an average speed over a period of time. The speed it reported was not real time.

View attachment 491253

Troubleshooting
Odd, but not uncommon. I've seen the same with a service called BackBlaze which I had also used in the past. The service sends data in chunks, waiting for the server to verify the integrity of the chunk before requesting the rest of the file. I decided to look into the gdrive-cli manpages and found that I was able to set the chunk size myself, keeping in mind this would mean the chunk be loaded into memory first which meant my system's memory usage would increase (slightly, on modern systems you probably wont be able to tell the difference)

I went for it, doubling the chunk size from it's default 8388608 to 16777216 bytes.

View attachment 491255

Pretty neat huh?

Comparison testing

Running with 8388608byte chunks (x1):
- Average speed over entire upload: 4.7MB/s
- Average memory usage change was negligible.

Running with 16777216byte chunks (x2):
- Average speed over entire upload: 7.3MB/s
- Average memory usage change was 0.01GB

Running with Running with 33554432byte chunks (x4):
- Average speed over entire upload: 9.9MB/s
- Average memory usage change was 0.01GB


Looks like we're hitting diminishing returns. With a x4 increase in chunk size I managed to more than double the speed of uploads. More optimisation is possible, but beyond the scope of this post.

https://mybroadband.co.za/forum/threads/my-review-of-cool-ideas-200mbit-vumatel.930392/page-2

Here are my findings doing testing with GDrive back when I was with CISP, this still holds true.

 

[Concentric]

Drool
That international

 

[Concentric]

https://mybroadband.co.za/forum/threads/my-review-of-cool-ideas-200mbit-vumatel.930392/page-2

Here are my findings doing testing with GDrive back when I was with CISP, this still holds true.

Thanks!
Im sure my europe server could thrash a 1Gps line with gdrive, although that was with rclone. I will test

 

[Mr Scratch]

am v happy

 

[cavedog]

am v happy

View attachment 637110

Ewwww not the Web rip there is a blu ray out already

Also I have been seeing more and more fibre ISP's in the peer lists of popular downloads. IS FTTH, Cool Ideas, Atomic, Infogro, MTN Business aka Supersonic and Macrolan. Then the usual Mweb,Voxdsl,Telkom, 8ta.Telkom Afrijost and Axxess. These should help you guys speed international torrents a bit.

Once downloaded a torrent from a Cool Ideas IP in bitcomet at 12mb/s so some people just leave their torrents uploading full speed ahead.

 

[DarkSt0rm]

Thanks!
Im sure my europe server could thrash a 1Gps line with gdrive, although that was with rclone. I will test

Tag me if you share the results, I use rclone with gdrive more often than the official client

 

[Mr Scratch]

Ewwww not the Web rip there is a blu ray out already

Jammer