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.
Comments
from my Datashack server to Catalyst Host in Dallas, TX
We have 10 gbps for VPSes with prometeus/iperweb and when people go to 800 mbps it starts affecting the node, sometimes faster if the pps is high so we consider abuse any sustained transfer over 500 mbps.
Usually, at 300 mbps and below we have no problems, the port is always far from saturated, right now we have 0.25 Gbps down and 0.35 Gbps up from all VPSes and it gets as high as triple during peaks, however, speedtests dont do great unless you test it close.
With torrents, many transfers, 800 mbps are easy to achieve, this is why we are keeping an eye out for this as it can spiral out of control.
If it's a VPS on a server 1Gbit port, then a client should really except a few hundred mbp/s, in my experience.
Well this is a colo server not a vps.
SO the consensus here is that if you aren't paying for a gigabit commit, and you are paying for something else (like 100TB) that you shouldn't be able to burst a gigabit pipe even if you have one........ that's really sad. If you have a gigabit commit you would probably be on a 10 gigabit port anyway or you wouldn't be able to use your commit of 1gbit. So then I guess people think you shouldn't be able to burst your 10 gigabit either?
No, not at all. You are definitely getting what you are paying for. Your DC does not give you a 10Gbit drop anyway. They give the SWITCH a 10Gbit drop, and provide you with a 1GbE network drop. No way in HELL would they give you your own 10Gbit drop unless you pay for it. Your logic here is really lacking.
It costs your DC money to "allocate" you 320Mbps even. Keep in mind you are SHARING a 10Gbit pipe between generally a full rack. So say you have 30 servers in a rack that are actively using the 10Gbit pipe, it is expected to see right around 320Mbps...
You get 100TB on a 1Gbit port, so theoretically, if those 30 other clients are only using 9Gbit/10Gbit, you could get that FD Gbit connection.
You want a FD 1Gbit dedicated port? You'll be paying for a cross connect to one of the providers that have a PoP in your DC.
Regardless if it is a colocated server, dedicated server, or VPS.
TL;DR: it's all based on availability, what speeds are available, you will see.
@DestroyeRCo that's the norm, i assume you was in KC.
When we was back in manchester, our switch had a gig uplink and we could burst it fine.
A good test might be to try something like iperf. I did some gbit tests at my home (at a previous house) and achieved 930-950mbit. That was just between two boxes over a gigabit link. Your milage may vary as you put many, many more factors in between on the internet.
Checking your speed between various vps providers anyway with iperf might prove helpful if you're concerned about your bandwidth.
I dont think you understand the difference in a commit and a port.
A commit is something I am billed for like 25mbit 95th percentile. Port speed is usually not billed for and is your 'burstable' rate.
A commit is something I am billed for like 25mbit 95th percentile. Port speed is usually not billed for and is your 'burstable' rate.
No, I obviously know that. Considering you are seeing 320Mbps, I am going to assume that you have xTB on a Gigabit port, not a xMbit commit. In which case, you are definitely getting what you are paying for on a shared gig port. 320Mbps is a very decent speed on a shared port.
If you do have a 1000Mbit commit, definitely let them know, because there would be an issue in that case.
There's also the issue with tcp window and high latency reduces the amount of throughput per connection. Unless it's a UDP test or you've tweaked TCP buffers on both server and client, it's not going to be definitive.
Why would they sell everyone on one switch on a shared gigabit port? I would think they would have 10gbit to the switch.
10Gbit -> switch 24 x 1Gbit -> customer servers
XTB vs xMbit doesn't make a difference they can be translated one into the other so I don't know what you mean. Are you saying if I had a Mbit commit I wouldn't be getting what I'm paying for?
I think the whole ' you are getting what you are paying for ' mantra is a little dumb.
I was told I was on a gigabit port(nothing said about sharing), I expect to be able to burst the gigabit.... why should I have to pay more or less? This isn't a VPS.
Could probably fit a lot more than 24 1gbit without it being oversold.
Err well 10 / 24 is .400 so 400Mbps for each client would technically be over subscribing a 10Gbit line, but that is the norm and to be expected.
You were not told you were on a dedicated port, hence you must be in a shared one. Basic logic
320mbit would be fine on a gig port for me
+1, that's what I've been trying to explain
Yep, typical shared gig speed
Because bandwidth costs money. You're essentially paying for a 300Mbps commit which you are getting better than.
Best I've seen in my mix of colo and rented VPS nodes is:
105M/s
Those speeds and a bit more - up to about 112M/s I see out of Kansas City and Dallas.
That's transferring real files, large file and often compressed beforehand.
I have not been guaranteed a gigabit port to myself... I think you are all confused..... it's 30TB on a gigabit port.
You're the only one confused here, thinking you should be getting "999Mbps" on a shared gigabit port. @KuJoe only said it was essentially like a 300Mbit commit, because you typically get right around 300Mbps on a shared gigabit port.
+1. This. Then when you are on a dedicated gigabit port you are also essentially on a shared port of something larger. It's never fully fully dedicated in most cases.
I use EdgeCast: http://c.speedtest.net/mini/mini.zip (Oh, sorry, ookla) They usually perform better than CacheFly.
Yes I was confused. You guys got it all straightened out for me. Thanks!
This thread opened my eyes also. After reading it, something clicked in my head and it all made sense to me.
Well, to be honest, from a 1 Gbit I expect atleast that I can max out 100 Mbit/s.
However this is what I got with my LoomHosts VPS for wich I vouch!
Download speed from CacheFly: 83.5MB/s
Download speed from Linode, Atlanta GA: 9.60MB/s
Download speed from Linode, Dallas, TX: 109MB/s
Download speed from Linode, Tokyo, JP: 4.91MB/s
Download speed from Linode, London, UK: 3.54MB/s
Download speed from Leaseweb, Haarlem, NL: 9.93MB/s
Download speed from Softlayer, Singapore: 5.13MB/s
Download speed from Softlayer, Seattle, WA: 8.52MB/s
Download speed from Softlayer, San Jose, CA: 740KB/s
Download speed from Softlayer, Washington, DC: 11.1MB/s
@Nevil based on the Cachefly test, that's 668Mbps.
However, I wouldn't count on Cachefly to base your speed on (:
STOP USING CACHEFLY.
Their speedtest is 90%+ fraud empty space. It has no indication of line speed. None.
How about the one from linode tx? Pretty decent if you ask me.