Cool Ideas Fibre ISP - Feedback

Status
Not open for further replies.
jannier said:
mmmm that is what I thought when I saw the latency, Why does everything always route to Johannesburg? Cost implications of having a presence in the major cities to reduce latency too High?
Click to expand...

A CDN doesn't need 1ms latency, it just needs bandwidth. No need to have them scattered around the country unless it's more expensive to provision the bandwidth than the service...
 
Sinbad said:
A CDN doesn't need 1ms latency, it just needs bandwidth. No need to have them scattered around the country unless it's more expensive to provision the bandwidth than the service...
Click to expand...

OH. I noticed that there is a option under the download section in the steam app to choose Cape Town instead of Johannesburg to DL from.
 
PBCool said:
MTN Customers in Cape Town, are first routed over MTNs longhaul to our JHB NNI, then to the rest of the network. Instead of local hand off in Cape Town.
Click to expand...

ah .. that would explain why even my very first hop (after router) is 20ms?

traceroute to googlemail.l.google.com (216.58.223.5), 64 hops max
1 10.23.23.1 0.411ms 0.491ms 0.297ms
2 154.0.14.172 22.781ms 22.068ms 22.742ms
3 154.0.4.30 23.090ms 22.577ms 22.127ms
4 154.0.3.82 22.930ms 22.848ms 23.117ms
5 154.0.3.205 22.712ms 22.555ms 22.186ms
6 154.0.4.82 23.019ms 22.654ms 23.131ms
7 196.60.8.166 24.673ms 22.901ms 23.026ms
8 72.14.237.239 23.058ms 22.684ms 22.936ms
9 216.58.223.5 22.799ms 23.011ms 26.766ms
 
blunt said:
ah .. that would explain why even my very first hop (after router) is 20ms?

traceroute to googlemail.l.google.com (216.58.223.5), 64 hops max
1 10.23.23.1 0.411ms 0.491ms 0.297ms
2 154.0.14.172 22.781ms 22.068ms 22.742ms
3 154.0.4.30 23.090ms 22.577ms 22.127ms
4 154.0.3.82 22.930ms 22.848ms 23.117ms
5 154.0.3.205 22.712ms 22.555ms 22.186ms
6 154.0.4.82 23.019ms 22.654ms 23.131ms
7 196.60.8.166 24.673ms 22.901ms 23.026ms
8 72.14.237.239 23.058ms 22.684ms 22.936ms
9 216.58.223.5 22.799ms 23.011ms 26.766ms
Click to expand...

Yeap, but should be resolved pretty quickly, waiting on MTN to patch their side, then it's just configuration and traffic will shift.
 
Good afternoon. I am moving to CISP at the end of the month on Open Serve infrastructure. It looks like Vumatel will also be installing in my complex. Would it be an issue to cancel Open Serve with CISP if I move over to Vuma infrastructure once ready?
 
Johan R said:
Good afternoon. I am moving to CISP at the end of the month on Open Serve infrastructure. It looks like Vumatel will also be installing in my complex. Would it be an issue to cancel Open Serve with CISP if I move over to Vuma infrastructure once ready?
Click to expand...
Nope, no problem.
 
Hey PB,

How do i go about upgrading my service 50/50 FH.

There use to be an upgrade option on my portal, but i cant seem to find it now.

Oh and a side note. I know a few pages ago there was talk around twitch and the inability to stream. I know there is a local youtube server (or something there abouts) and i am able to stream live games at 1080p with no issue...

What do we need to do, to get twitch to host a server here?
 
Bedford said:
Hey PB,

How do i go about upgrading my service 50/50 FH.

There use to be an upgrade option on my portal, but i cant seem to find it now.

Oh and a side note. I know a few pages ago there was talk around twitch and the inability to stream. I know there is a local youtube server (or something there abouts) and i am able to stream live games at 1080p with no issue...

What do we need to do, to get twitch to host a server here?
Click to expand...
FH as in Fibrehoods? This is a manual process so please email [email protected] with the request.
 
WRT twitch we have offered to host and provide transit for them, they aren't interested. So not sure to be honest.
 
PBCool said:
WRT twitch we have offered to host and provide transit for them, they aren't interested. So not sure to be honest.
Click to expand...

At this point in time i am using a Local VPS to act as a relay to Twitch...

So i send my data to the VPS and it sends it along to Twitch.

I have found that using this method has resulted in no drops of frames or throughput issues :)

The Server uses minimal hardware resources (1gig ram and 1 CPU core) the only thing is requires is a beefy bandwidth connection.
 
DJZassie said:
At this point in time i am using a Local VPS to act as a relay to Twitch...

So i send my data to the VPS and it sends it along to Twitch.

I have found that using this method has resulted in no drops of frames or throughput issues :)

The Server uses minimal hardware resources (1gig ram and 1 CPU core) the only thing is requires is a beefy bandwidth connection.
Click to expand...

What OS is running on the VPS?
 
PBCool said:
What OS is running on the VPS?
Click to expand...

Ubuntu 16.04

Currently using DataKeepers ... 1ms ping to the server in CPT ^.^

Where i would normally get around 20,000 odd dropped frames (3%-4%) in a 3-4 hour stream, i now drop nothing at all
 
Yeah so this is because the Ubuntu box is doing the TCP scaling for you to the international destinations, your PC to the VPS doesnt have to due to the low latency. If you ran the VPS on your LAN you would probably have the same result. Also how are you routing your twitch traffic via the VPS? VPN or proxy?
 
Last edited:
PBCool said:
Yeah so this is because the Ubuntu box is doing the TCP scaling for you to the international destinations, your PC to the VPS doesnt have to due to the low latency. If you ran the VPS on your LAN you would probably have the same result. Also how are you routing your twitch traffic via the VPS? VPN or proxy?
Click to expand...

I agree, the no drop in frames on my end is exactly because it is a local connection.

This isn't to Watch twitch, this is to STREAM TO TWITCH... aka i am not downloading any data from twitch but rather uploading data to twitch.

To upload to twitch it uses the RTMP protocol, what my server does is simply relay the traffic i send it to the twitch ingest servers in London.
I have Nginx with the RTMP plugin installed on the server.
the server has an open port for the RTMP traffic, i then obviously send the data to the server IP address and port from my capture software.
The server then forwards the stream of data to the twitch ingest servers using the server providers network :)

So effectively the server is acting as an ingest server for RTMP traffic. Once it receives an RTMP stream of data it then does something with it, in this case it just forwards it onto the twitch ingest server, which then gets distributed over twitch :)

Basically my server acts as a middle man because it has a stronger/more stable connection to EU than my residential connection .
 
DJZassie said:
I agree, the no drop in frames on my end is exactly because it is a local connection.

This isn't to Watch twitch, this is to STREAM TO TWITCH... aka i am not downloading any data from twitch but rather uploading data to twitch.

To upload to twitch it uses the RTMP protocol, what my server does is simply relay the traffic i send it to the twitch ingest servers in London.
I have Nginx with the RTMP plugin installed on the server.
the server has an open port for the RTMP traffic, i then obviously send the data to the server IP address and port from my capture software.
The server then forwards the stream of data to the twitch ingest servers using the server providers network :)

So effectively the server is acting as an ingest server for RTMP traffic. Once it receives an RTMP stream of data it then does something with it, in this case it just forwards it onto the twitch ingest server, which then gets distributed over twitch :)

Basically my server acts as a middle man because it has a stronger/more stable connection to EU than my residential connection .
Click to expand...

Yeah so as mentioned this is probably only working because of the outbound TCP window scaling the linux kernel is doing for you, if you ran the server on your LAN you would probably have the same result.

This has nothing to do with the "strength or stability" of the hosted connection.

We could host one of these on our core for all twitch uploads if it's something people would possibly want?
 
PBCool said:
Yeah so as mentioned this is probably only working because of the outbound TCP window scaling, if you ran the server on your LAN you would probably have the same result.

This has nothing to do with the "strength or stability" of the hosted connection.
Click to expand...

I'll load up Ubuntu on one of my spare machines back at home and give it a test ^.^

Pitty Windows Auto-scaling is so bad
 
Status
Not open for further replies.
Back
Top
Sign up to the MyBroadband newsletter
X