Afrihost congestion (on OpenServe) getting worse

Deckert

Deckert

Well-Known Member
Joined
Jan 13, 2004
Messages
425
Reaction score
38
Location
Centurion, South Africa
hi,

I've had my 20Mbit Afrihost connection since October 2017. I've been exceedingly happy with it, but things are starting to deteriorate. I've been monitoring my link latency since Jan 2018. It's been pretty good, but April is when the problems started.

See the graph below for a trend showing *average* ping times getting worse:

afrihost-trend-2018.jpg

If you zoom in to a 30-hour period, the congestion issue is quite clear. Between about 18:30 and 22:15, throughput plummets and latency increases dramatically, from a pleasant 6ms all the way to 130ms.

afrihost-pings-20180706.jpg

Ping tests are done once every 5 minutes with 20 packets.

--deckert
 
Deckert said:
hi,

I've had my 20Mbit Afrihost connection since October 2017. I've been exceedingly happy with it, but things are starting to deteriorate. I've been monitoring my link latency since Jan 2018. It's been pretty good, but April is when the problems started.

See the graph below for a trend showing *average* ping times getting worse:

View attachment 534493

If you zoom in to a 30-hour period, the congestion issue is quite clear. Between about 18:30 and 22:15, throughput plummets and latency increases dramatically, from a pleasant 6ms all the way to 130ms.

View attachment 534495

Ping tests are done once every 5 minutes with 20 packets.

--deckert
Click to expand...

It's difficult to diagnose from the smokeping - depends on how you've setup the tests and where your server is located.

It might be easier for us to see a local and international traceroute, as well as a ping test to bras.afrihost.com. Since you are using Openserve, it should still run through their local infrastructure.
 
The same thing is happening to Axxess as well. Congestion in the evenings renders it almost unusable.

What software are you using, I would love to run the same tests?
 
Deckert said:
Not following your question. You referring to what I'm pinging?

--deckert
Click to expand...

Yes.

bras.afrihost.com is the equipment that sits just behind the exchange. Since the exchange doesn't respond to the ping/tracert protocol anymore (OpenServe was worried that it might lead to DoS attacks :wtf:), you can ping bras.afrihost.com which is just afrihost setting up an easy domain to remember 155.239.255.250

If that has higher latency, it isn't your ISP but OpenServe/exchange.

EDIT:
You can also ping it from another ISP, e.g. I am currently on Vox.
Code: 
Pinging bras.afrihost.com [155.239.255.250] with 32 bytes of data:
Reply from 155.239.255.250: bytes=32 time=2ms TTL=63
Reply from 155.239.255.250: bytes=32 time=3ms TTL=63
Reply from 155.239.255.250: bytes=32 time=2ms TTL=63
Reply from 155.239.255.250: bytes=32 time=1ms TTL=63
Reply from 155.239.255.250: bytes=32 time=3ms TTL=63
Reply from 155.239.255.250: bytes=32 time=2ms TTL=63
Reply from 155.239.255.250: bytes=32 time=4ms TTL=63
Reply from 155.239.255.250: bytes=32 time=3ms TTL=63
Reply from 155.239.255.250: bytes=32 time=1ms TTL=63
Reply from 155.239.255.250: bytes=32 time=1ms TTL=63
Reply from 155.239.255.250: bytes=32 time=3ms TTL=63
Reply from 155.239.255.250: bytes=32 time=2ms TTL=63
Reply from 155.239.255.250: bytes=32 time=3ms TTL=63
Reply from 155.239.255.250: bytes=32 time=1ms TTL=63
Reply from 155.239.255.250: bytes=32 time=3ms TTL=63
Reply from 155.239.255.250: bytes=32 time=2ms TTL=63
Reply from 155.239.255.250: bytes=32 time=1ms TTL=63
Reply from 155.239.255.250: bytes=32 time=3ms TTL=63
Reply from 155.239.255.250: bytes=32 time=2ms TTL=63
Reply from 155.239.255.250: bytes=32 time=1ms TTL=63

Ping statistics for 155.239.255.250:
    Packets: Sent = 20, Received = 20, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 1ms, Maximum = 4ms, Average = 2ms

Though mine is slightly high in jitter as over WiFi. (I'm on Fiber)
 
Also having the same issues during peak time on my Afrihost 10MBps (Openserve), getting around 2-3 down and 4 up. Ping sits around 100-150 on the mybb speedtest.
 
I've had terrible pings and non-useable GitHub downloads between 6pm and 10pm the last few months. Even went to CAPPED Afrihost data package, and still terrible. Afrihost just denied there was an issue.

Finally gave up and now rocking on Cool Ideas - no more problems.
 
Johnatan56 said:
bras.afrihost.com is the equipment that sits just behind the exchange. Since the exchange doesn't respond to the ping/tracert protocol anymore (OpenServe was worried that it might lead to DoS attacks :wtf:), you can ping bras.afrihost.com which is just afrihost setting up an easy domain to remember 155.239.255.250

If that has higher latency, it isn't your ISP but OpenServe/exchange.
Click to expand...

Now that is useful information - thank you.

So let's find out if it's OpenServe or Afrihost. However, even if it is OpenServe, I would expect Afrihost - and any other ISP for that matter - to have some sort of minimum SLA with OpenServe. It's a bit too late now, but I'll setup smokeping from the inside and ping bras.afrihost.com.

I'll also run some regular mtr reporting scripts at set intervals "just in case". I'll let it run a few days and report back in this thread.

Code: 
$ date
Fri Jul  6 21:46:18 SAST 2018

$ ping bras.afrihost.com
PING bras.afrihost.com (155.239.255.250) 56(84) bytes of data.
64 bytes from 155.239.255.250: icmp_seq=1 ttl=63 time=2.56 ms
64 bytes from 155.239.255.250: icmp_seq=2 ttl=63 time=2.48 ms
64 bytes from 155.239.255.250: icmp_seq=3 ttl=63 time=2.73 ms
64 bytes from 155.239.255.250: icmp_seq=4 ttl=63 time=2.59 ms
64 bytes from 155.239.255.250: icmp_seq=5 ttl=63 time=2.69 ms
64 bytes from 155.239.255.250: icmp_seq=6 ttl=63 time=2.52 ms
--- bras.afrihost.com ping statistics ---
6 packets transmitted, 6 received, 0% packet loss, time 5007ms
rtt min/avg/max/mdev = 2.485/2.601/2.737/0.098 ms

--deckert
 
Deckert said:
Now that is useful information - thank you.

So let's find out if it's OpenServe or Afrihost. However, even if it is OpenServe, I would expect Afrihost - and any other ISP for that matter - to have some sort of minimum SLA with OpenServe. It's a bit too late now, but I'll setup smokeping from the inside and ping bras.afrihost.com.

I'll also run some regular mtr reporting scripts at set intervals "just in case". I'll let it run a few days and report back in this thread.

Code: 
$ date
Fri Jul  6 21:46:18 SAST 2018

$ ping bras.afrihost.com
PING bras.afrihost.com (155.239.255.250) 56(84) bytes of data.
64 bytes from 155.239.255.250: icmp_seq=1 ttl=63 time=2.56 ms
64 bytes from 155.239.255.250: icmp_seq=2 ttl=63 time=2.48 ms
64 bytes from 155.239.255.250: icmp_seq=3 ttl=63 time=2.73 ms
64 bytes from 155.239.255.250: icmp_seq=4 ttl=63 time=2.59 ms
64 bytes from 155.239.255.250: icmp_seq=5 ttl=63 time=2.69 ms
64 bytes from 155.239.255.250: icmp_seq=6 ttl=63 time=2.52 ms
--- bras.afrihost.com ping statistics ---
6 packets transmitted, 6 received, 0% packet loss, time 5007ms
rtt min/avg/max/mdev = 2.485/2.601/2.737/0.098 ms

--deckert
Click to expand...

Lol, that SLA doesn't count, near always it has been due to Afrihost not upgrading the amount of IPC, which is why it happens at peak time. It's why I left Afrihost after the nth time it happened, just got so annoyed, plus my experience of their critical care is that it is useless.

I suggest you try another ISP account and see if the problem persists.
 
AfriGenie said:
How are the results looking so far?
Click to expand...

Here's a comparative plot between ping times to bras.afrihost.com and 1.1.1.1 (a local cloudflare DNS server). The probe I wrote sends 10 ICMP packets of 56 bytes every 60 seconds. It pings bras first then the cloudflare server directly afterward.

afrihost-bras-ping-2018-07-09.jpg

I'll post today's plot once the day finishes. So ping times to bras.afrihost.com is basically pretty stable, but the ping times to quad-one fluctuate and increases during the evening time peak.

--deckert
 
Deckert,

Customer sent me this... Afrihost / Openserve Fiber. Same story every night.

customer2.jpg
 
mister said:
Deckert,

Customer sent me this... Afrihost / Openserve Fiber. Same story every night.
Click to expand...

Deckert said:
Here's a comparative plot between ping times to bras.afrihost.com and 1.1.1.1 (a local cloudflare DNS server). The probe I wrote sends 10 ICMP packets of 56 bytes every 60 seconds. It pings bras first then the cloudflare server directly afterward.

I'll post today's plot once the day finishes. So ping times to bras.afrihost.com is basically pretty stable, but the ping times to quad-one fluctuate and increases during the evening time peak.

--deckert
Click to expand...

It definitely doesn't look good - but our NOC report that on our side the network is well within usage limits and we're currently doing minimal shaping - even at peak.

Can you please post the full traceroute so they can analyse where latency originates?
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X