All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
help a noob understanding his connectivity
Disclaimer: not a pro, not at all, just a hobbyist, so be gentle 
. I might ask stupid basic questions, I haven't studied this shit
I do have some servers around, couple of idlers NATted ones, a seedbox on scaleway (sorry guys, I will move it to one of your smaller business soon tho, even if it's just a few bucks I do want to support smaller companies) and a few yearly ones around Europe (first-root, gullos, etc.)
1) what tools should I use to understand the actual connectivity?
Right now I'm looking at traceroutes, ping time and I use iperf to measure bandwidth amongst those, I can write simple script and then use simple tools (python) to plot the results, I've recently learned about AS numbers and BGP routing and I'm adding the AS number to my observations, right now I just lookup the AS from the IP, including the IPs I see popping up in the traceroutes, and by looking at the graphs I make I try to make sort of a mental map of what's been routed where.
what tools do you guys use to do this and does this makes any sense at all? I mean am I getting a real picture of what's happening to my packets during normal use?
Say for example that I noticed that my packets towards a VPS in Paris get routed through marseille and then Paris on the traceroute output, and iperf is telling me I get 7/8 MiB at the same time on the same hosts, other packets towards VPS in Germany were routed through Milan and Dusseldorf and I get 5 MiB,
Can I do this? can I assume that the IPs I see in the traceroute are more or less the same routes that all the other packets are taking too? Can I assume that if I get those measurement from iperf I can then use those to monitor connectivity and get at least an Idea of what I'll be getting from and to this or that host?
What about speedtest-cli? is this supposed to be a reliable tool? because If I measure the bandwith on my servers with that I'm getting very different numbers than if I measure the bandwidth between the hosts on my network.
and what about the actual network performance? the only network heavy services I do are streaming from a plex server and backups, when I do that I see much higher throughput, like 20/30 Mib, which actually is what I'm supposed to have at home.
so how do I reconcile those numbers? why do Iperf provides such different numbers than reality?
2) what I want to play with (again, this is just an hobby for me) is getting the fastest possible connections between my home (north west of Italy) and the rest of my network, everything runs through VPNs by the way, one of those yearly dedi is a pfsense box I use kinda like a VPN aggregator.
this small provider I'm using (Linkem) has an AS number which is only registered at an exchange in Milan tho, so should I just get a VPS in Milan and VPN Into that and then have my packets being routed from there? would this make sense and improve my connectivity to the rest of my network just because it makes sense geographically?
I know that peering agreements are not public, and especially when it comes to smaller providers who knows, but If I see the asnumber of my home connection only on a single exchange can I assume that if I get a VPS in Milan who's network AS number is registered on the same exchange I'll "improve" somehow my connectivity?
See I do have a 30Mbps connection (speedtest from my home on their websites), the fact that I'm only getting 7/8MiB at tops and usually just around 5 to most providers here (but that applies to bigger guys too, like hetzner or scaleway) doesn't really add up.
The fact then that I'm actually getting more or less the same numbers from most of the host during actual usage is even more confusing to me, because taking those measurements then doesn't seem to make much sense, like any VPS anywhere (at least in Europe) seems to top my home connections during actual usage but not during measurement with iperf.
all of those numbers are using the VPN, I also do the test without and I consistently see a pretty small overhead, I get an average of 0,5/1 M more without the VPN.
thanks in advance for helping me understand!
I love shopping for servers here and have fun with them, makes me feel accomplished when everything works and I then leave it there idling, so thanks for the endless fun LET
Comments
Hi @dada216 216
Speedtest can give a rough number on how fast your Internet connection is. A speedtest depends on multiple factors: Where is the Server located, in which Network, is it connected with a single Gbit Link or a 10Gbit Link?
I have also seen providers limiting speedtests in generall as it's bandwidth heavy and does not carry any actual content.
Yes, however, packets can go a different route from a Server to you than from you to the Server.
Linkem is connected to the NaMeX in Rome, not Milan
Looking at your ISP, no i don't think so. Your ISP is connected to that IX but it looks like they don't do any peering there. They are even listed as
not peeringon the IXP Website..If a provider offers services in a city in which your ISP is present in, does not mean you get a good and fast connection.
I think you are best off getting a VPS with a provider that has Cogent, GTT or TATA as their Upstream, as your ISP uses them as well.
Well, it does. As speedtest takes the "nearest" Server, its probably the speedtest server in your provides network, to which your Provider will maintain good connectivity.
It would be interesting to see MTRs to those Hosts as you can see packet loss with it as well.
I agree with the suggestions by @ruben, but also please can check the units you're giving for speeds, because 30 Mbps (megabits per second) is equal to 3.58 MiBps (mebibytes per second), so a 30Mbps connection would probably explain what's going on.