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
with @jarland I got 60MB/s Up and down. with @Jacob I got 60MB/s download and 30MB/s upload. Can't remember any other xD
If I could get 320mbit would be great. A lot of providers don't even get close to that.
Is this on average (and, depending on network/node conditions, can get even more?) or just limited to 320Mbit?
Well I have a colo provider telling me that 320mbit from cachefly is GREAT and this is my first gigabit port... so i'm not sure. I figured I would be able to burst 999mbit to almost anywhere. Especially cachefly.
Cachefly is not a good measure of network speed.
what is? 100 connections to random websites?
Nah, not unless you get a dedicated Gbit port, which you will specifically know you got considering the price of one.
I agree with KuJoe
Can get 60MBps from cachefly on my datashack dedi, but can get 118MBps from softlayer dallas.
Cachefly is never more than 50MB/s for us and their compressed file is a bit deceptive and I'm not a fan of their throttling. BurstNET, GigeNET, and RamNode offer much faster speeds than Cachefly does for us.
Unless they say dedicated gig port you wont even get close to full capacity. And a dedicated gig port it usually around $800+
Personally, on all the gbit port vpses I've had, I've only been able to get like 10mb/s.
Remember, there is the middle and far end of the path too. Regardless of how much bandwidth your host can support, you'll never transfer faster than the intermediate and far-end links can support. Generally speaking, multi-threaded tests taking various paths are a good way to get a feel.
On a VPS the port is shared, and there is no guarantee the host's uplinks can support such speeds. Gigabit ports relieve congestion above 100Mbps, but they don't necessarily imply actual gigabit speeds.
This is a colo server with a gigabit port. I expect near gigabit speeds. It is not and should not be shared - if it were I might as well go get a 100mbit port from someone else. My commit is not 1gbit and it shouldn't have to be to get a gigabit port. Saying a gigabit port alone is $800+ is silly when I only use a fraction in the commit.
Ok, gotcha. Worthwhile to make sure all the network settings are kosher and then do some multi-threaded testing. Middle and end still apply.
Keep in mind "gigabit" is anything > 100mbit. Basically what you get is a little slice of a 2-10Gig per rack setup usually. It's all fair share usage, and when the usage is low, you'll be seeing higher speeds, but when the network is seeing high usage, you'll obviously see lower speeds.
If you want true FD 1000Mbit, you'll be paying quite a bit for it. Generally anywhere from $.50/mbit to $3/mbit. Typically lower when you commit higher. For example, if you only commit 50Mbps, upstream providers generally won't cut you any deals because there wouldn't be much profit to be made.
And like stated above, even then you wouldn't see "999Mbps", due to network overhead, and all of that.
So less than half of that is acceptable?
A bit deceptive? Cachefly ought to be SUED for being misleading about their speed test FRAUD.
The file compresses by 90%+. Is that speed you get impressive for say a 10MB file or a 1MB file? Cause that is all you folks are really downloading.
Yep, for the price you pay, the speeds you are seeing are definitely within what you are paying for. If you're in need of more bandwidth, contact your DC and ask if they offer any cross connects to any of their upstream providers, most with do it for a small fee, some will do it for free. Or you could even just ask what their pricing would be on a 1000Mbit commit.
That is why I hate it when people treat Cachefly's speed as gospel. We always use GigeNET as a much better representation of our network true speeds (we use BurstNET and RamNode if we want to make sure our Gig uplink it working normally also).
Not to mention if you download the test more than once per hour for X amount of hours in a row they will throttle you quite a bit (less than 100Mbps). I think this is based on subnet and not individual IP because we have seen this happen in the past after we get posted on LEB and 40+ new orders decide to test Cachefly after they first login within minutes of each other.
I think Singlehop or someone did that as well. Pretty funny actually, I downloaded a 100MB file in 3 seconds on my home cable connection
But yeah, pretty fraudulent, don't let Cachefly fool you.
Might just be spoiled with incero, but I get pretty much full gigabit
Nobody is saying that you won't get gigabit speeds on a gigabit port, they are just saying if you're not paying for 1Gbps you aren't entitled to guaranteed 1Gbps speeds. In the majority of settings, network ports are shared ports. Even if your 1Gbps port is line rate, the switch uplink may only be 10Gbps and sharing 24 1Gbps ports with a 10Gbps uplink doesn't add up.
@KuJoe
Got it, however if I'm paying for it. I'd like to use it, well most of it
Not like I ever come close to using that much anyways lol
It never adds up, but that's why there's usually a cap, e.g. 100TB. It's again a balance of overselling vs overloading. Usually you put enough and balance it so everyone gets decent speeds.
Agreed. If you are paying for a 1Gbps commit then you better be able to hit 1Gbps at all times. But if you are paying for a 1Gbps port with 10TB of bandwidth then you're not exactly paying for a line rate 1Gbps port.
Also true, but I was working under the assumption that they were paying based on a commit (normally 95th percentile) as that's how most colocation is that I've seen.
It's similar. You can convert the commit back to TB/month anyway or similar. I bet they estimate both the same way.
I expect to be able to burst up to 800mbit on a perfect connection for at least a few seconds if the line isn't seeing hardly any use. I expect an average of 300-400mbit and a consistent ability to hit 100mbit to any location with a decently strong network on the same continent.
As a customer of a provider of course. As a customer of the upstream provider itself, I better get the advertised speed anytime the route is decent.
YESSS I hate that. I remember one day the same VPS was alternating between 50 and 4 MB/s from Cachefly with no change in node activity T_T
@Corey - when are you getting your next server there? If you're just trying to confirm your port speed, you can run tests between local servers. Probably obvious, but I figured I would at least mention it.
The best I have got is 101 MB/s for a gig port.That server was just deployed,so no site on it
From cachefly by using Windows, I can get 70mb/s with FrontRangeHosting, and 100-150mb/s with Prometeus. Compared also with speedtest