New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Exabytes vs. Smarthost.net Denver Routing within same datacenter is different
Hi,
I was looking at SmartHost Denver and Exabytes comparing the two providers, both located at Handy Networks Denver data center. Exabytes provides the test IP: 173.248.157.218. I also compared SmartHost's Denver test IP of 192.199.242.2. For some reason, despite both being located in the same data center, Exabytes has superior routing to most of our clients and lower latencies. I was wondering if anyone could shed any light on why this might be? Is it carrier related? How are these things usually figured out to make the right decision of what host to go with?
Thanked by 1jugganuts
Comments
Your probably better off with Chicago or Dallas in general.
192.199.242.2 is singlehomed zayo while 173.248.157.218 is zayo, gtt, cogent and he.
https://bgpview.io/prefix/173.248.128.0/18#routing
https://bgpview.io/prefix/192.199.242.0/24#routing
Testing both from the Netherlands, they seem practically the same from me. Both are almost entirely Zayo.
bgp map for the exabytes test ip: https://lg.ring.nlnog.net/prefix_bgpmap/lg01/ipv4?q=173.248.157.218
Exabytes has better network mix.
being at the same DC not always means same network...
Could you tell me how I could go about figuring this out in the future for choosing low latency VPSses?
You get yourself 2 or more servers to compare, with specific test destinations.
So you conclude tests and figure out which is better connected.
https://github.com/Ne00n/latency-triangulator-integrator-analysator-7560b
Before I add servers to my low latency network, I use that tool above.
I think the tool you describe is not ideal because I'm not just looking for myself, it's also for other clients from different locations. I am looking for carriers that have good reach to multiple states, not just for myself.
can try test IP using http://ping.pe/ ; http://ping.sx/ or https://dnstools.ws/
covers majority of locations, though actual local ISPs might route differently.
Thank you this is awesome!
One more question: what kind of "carrier mix" should I be looking for? Is there a rule of thumb? More carriers = better = lower latency? 1 carrier = worse = high latency? Is a provider with 5 carriers going to be slower than a provider with 10 carriers?
Is there anything I should look for in the following categories? Does any of this matter for latency from different states?
IPV4 PREFIXES: 55
IPV4 PEERS: 8
IPV4 UPSTREAMS: 5
IPV6 PREFIXES: 3
IPV6 PEERS: 7
IPV6 UPSTREAMS: 4
We can always tag Shawn with @SmartHost and see if maybe he can work some magic in terms of getting better routes.
We tried Chicago (RackNerd) and Dallas (Linode) it does not have better routing for our users. We also tried Kansas (Wholesale Internet) and again worse routing than Exabyte's. I am very curious how this all works and would love to understand it better.
LMAO
Try an actually good provider in Chicago, not some random ColoCrossing reseller crap.
@TNAHosting is pretty good, as well as Vultr (their Chicago location is amazing in terms of connectivity).
Here is where you can test Vultr:
https://il-us-ping.vultr.com/
And tnahosting:
http://lg.chi.tnahosting.net/
NFOServers also has a premium blend location in Chicago, and comes with a pretty good amount of resources for a pretty low price. ($9.99 for 12TB bandwidth, 2GB ram, 2 cores and 200GB storage).
Here is where you can test nfoservers:
https://www.nfoservers.com/networklocations.php
I think you're hitting the core of my question, which is how do I find providers which meet a requirement of "low latency." And it sounds like the way to do that is to look for "premium blend" of carriers? However I'm looking for a quantitative way of doing that. Is there a way of determining which VPS provider has a "premium blend" without relying on their promises and marketing-speech?
Use the testing links that I added as an edit to my post.
Also - never trust a providers word that they are the best. Just use the tools that they provide to do your own research, and determine if they are the best for you :-)
I'm not sure a simple traceroute helps as that does not tell me "premium carrier mix." It simply shows how I alone route to the server, not many different clients route to the server.
Ask your client to do mtr/traceroute the server IP from their network.
As @NobodyInteresting said the blend matters more than the location. Though most of the midwest such as Denver is routed through Dallas or Chicago. WSI uses mostly HE & Cogent and so does CC I believe. You could try giving NFOServers in Chicago a try.
That is a direct IP from them for testing, try one of ours instead and see if any better:
66.11.121.2
37.221.111.1
193.7.200.1
~ SMARTHOST
Not necessary always I will say, it depends & there are lot of variables at play.
Bumping this in case anyone else has advice on how to measure connectivity from multiple locations
Well, you can use ping.pe or RIPE Atlas.
https://mtr.sh
https://ping.sx
https://ping.pe
thanks. I'll test it for my ips
Actually this is an interesting topic.
For those providers here, how do you decide which or how many network carrier to use? is it dependent on the datacenter or cost or setup complexities? Is there a difference in monthly running costs?
I presume the bandwidth usage from your clients won't change regardless on who you select as carriers.
You read about Datacenter being carrier neutral, I presume that means each provider can choose their own carrier(s).
Turnkey internet based in Latham NY, has high latency to NYC but once they connect to NYIIX, the latency improved dramatically.