If you honestly think that a traceroute shows the truth path you are sadly misinformed, MPLS is very commonly used and hides several hops along the way.

Listen, I am not saying OVH is bad. Ovh is very large and well known and do a good job at what they do. The main reason OVH offers such pricing is because of their peering relationships with ISPS, yes they have a lot of peering, for things they don't peer they send out their transit lines. I.e bandwidth they buy. For the longest time ovh struggled to lots of places out side of the EU i.e US based ISPS. If you trace to someone like us, or quadranet or related due to peering, a network will send a peering partner their originating prefixes and their _customers_ prefixes. so in your first trace to quadranet you go OVH to mzima then to quadra, I"m pretty confident OVH and Mzima now known as Packet Exchange who is a very large EU network peer in.. wait for it.. Amsterdam in AMS-IX.


Hit end users in various networks in the states, Comcast, Cox, Paetec/Frontier/Cavtel etc. The eyeball networks. Hitting another hosting company is probably going to yield good speeds due to peering relationships.

Lets look at the states.


your 157ms vs my 65ms this can be a huge impact on lots of applications. While downloading a file between datacenters can be fast, lesser circuits may not be. I am just saying you really should be careful about hosting and place your content where it best suits your target audience, because in the end of the day latency matters.

*that* is exactly my point right from the start. I know I can get good traceroutes from ovh to other DCs because of the exact reasons you stated. However, the reverse is NOT always true. I have servers in QN and tried the reverse and it went over cogent and all over the frikken place. Hence why my download speeds from that server are shit. And they will soon be moved to ovh.

That is *exactly* why I gave that test doc as a download test (just happens to be a frikken excellent magazine too) to see IF US people could get good speeds.

For the record, that QN traceroute and download only worked in that direction - if I showed the other way it would be frikken awful. But I am not out on a smear campaign - I am purely trying to let people know of the others available (and maybe take in some admin'ing duties in the meantime).

Like people have mentioned in this thread (yourself included), people simply say 'Impossible for OVH to offer those prices with quality service, their transit sucks to the US etc etc" and it's well, just bullshit. That is what I am debunking.

OVH have had their problems, sure like every other DC, but shit, what they've accomplished in 12 years - they get a hat off to them. I took chrage of my first dedicated server in 1996. I have gone from host to host to host since then (always US, once UK2.net) and now OVH. AND I can tell you, I have *never* felt at home in any other DC as I have at OVH. They give me, a server admin, every single tool I could need to monitor, problem solve, problem detect. Never before have I had that amount of control over my server. And that is EXACTLY what I want. Now they give me the same control over telephones and my personal ADSL/SDSL line. I am for once fully in control of everything I do on the internet, and it's thanks to them. So I'm sending them some free pub. Shit, it did turn into a thread high jack.

You really have no idea how the whole network/pop game works do you? Just because someone has 28 Pops and they peer in 15 of them does not mean that those 15 IX points don't have a lot more capacity than the other 13 POPs.

Lets look at their peeringdb entry.

http://www.yellowfiber.net/clients/S...59.26%20PM.png
and
http://www.yellowfiber.net/clients/S...59.39%20PM.png

Now this is just public peering capacity at the given points not to mention "private" peering they have as well that is undisclosed. If you REALLY think they have MORE capacity to their paid transit lines than what they have in peering then you really need to read more about how networks actually operate. Ovh is also known for saturating circuits and not caring about it. Its pretty clear that 4pm EST on a saturday is not prime time.

Ovh is not the end-all/king of everything so please get off your high horse, they do a great job yes but they are not the total solution. IIRC they wont even sell a server to someone in the states and their policy was no one could resell their service either.

As I don't have a US ADSL line, can all US (hey, canada+world too) people tell me their download speeds on this file from two different servers:

NOTE: this download must be your home/office connection, NOT your server!!

1: http://204.152.194.186/tmp/OVH-MAG-2010.pdf

2: http://178.33.218.112/tmp/OVH-MAG-2010.pdf

Well, this is about the bottom line, since the rest is all chinese for me. All I know is:

1. none of my B2B customers have even uttered a breath since I switched to OVH. Hell, I don't think they even noticed. And I have 16 million daily B2B hits. When shit slows down (like it did in mid-Jan when I tried their cloud setup), I heard about it, so my customers are not simply not alert.

2. Google - my site has dropped from an average 4 second load time to 1.1 second load time since I switched.

3. My B2B traffic went up from 5mill to 8 mill in July last year because of some changes I made. In Dec when I moved to OVH, and I shit you not, it jumped to 12 mill almost from the day I switched DNS over to the new server. WTF that was all about I've no idea (my code tracking hits).

4. My bandwidth was a steady 4.3 Mbs at the switch at my old host. When I switched it dropped to a steady 1.1 Mbs at the switch. Ehm, same traffic, actually nearly 4million hits more per day yet nearly 4x less bandwidth. WTF that all about?

5. I can now frikken *do* shit with my stuff. I need geo-ips, they give them to me, I need a virtual bay of 4 servers (3 diff DCs), they give that to me.

I couldn't give a rat about connectivity if me, in my office sees a massive, massive drop in headaches following a DC move. I've maxed out my ADSL line on download tests with every single US host I've been with, only to find they were only ever rectified whenever I moaned. Guaranteed every single one went back to <50% speed in under 6 months.


Of course, OVH isn't for everyone, I'm just letting people know they are frikken good and a lot better a ISP than I have ever known in my life. If your customers get lousy speeds with them, they are not for you. I also heard by their mailing lists, they are building a DC in the US... *cough*

Freedom of choice is what it's all about.

Of course they're known for it - they even say it themselves :1orglaugh:1orglaugh

If you order their entry level Kimsumfi range and get 100Mbs they have said on their forums (or in their mailing lists, but I could dig it out), you will not get the same access as their HG servers which have 40Tb bandwidth and an own infrastructure dedicated to them. That is what their SLA Standard, SLA Premium, SLA business bandwidth levels are all about....

It isn't hidden, so don't make out like it's something they try and hide.


-edit on this point:

can you give me a URL where I can see this so I can look around?

On those grabs, there are checkboxes checked that you can uncheck - what's that all about?

http://www.ovh.ie

was setup to sell servers to the US and Canada (shit, anyone not in the UK and mainland Europe really - no vat)

And yes, I resell and have a customer support ticket stating that while they don't have a reseller program, they allow clients to resell under their nichandle. Under those terms, you are liable for all etc etc, and the disclaimer for reselling, you would need xyz because bandwidth is grouped under normal non-pro setups. ie 40Tb is per infrastructure, so to resell them, you'd need xyz to be able to offer the same independently.

So, yes, you are wrong. Problem is being a french company, if you don't speak french, you may find it hard to communicate (only 11 of their 300 team are native english speaking). Can I resell you some stuff for yellowfiber?

I have a 100M fiber connection at my house through my HOA but the first one was 890KB/s the seconed one is about 600KB/s * 8 and you will get your Mbps

are those averages as output from wget/fetch/cURL?

for a 100Mbs fiber connection, I'd expect more from both.

to add to the record, on a 28Mbs ADSl line (but because of attenuation, only 15Mbs can be max in thory), I get (a wget avg) of 610K on the first and 1.1MB on the second.

So the lesson from this is Spudstr from YellowFiber concedes that if you run a massive tube network (you know, those frikken massive TB-sucking content things) and deliver it over CDN for best end-user speeds, you'd be better off hosting with OVH cos their prices are shit low, their DC connectivity is great and well, the CDN takes over the last mile to deliver shit fast content delivery to your customers.

Hell, in that situation the increased costs in implementing a quality CDN are covered by the decreased costs in switching ISP. :thumbsup

Tracing OVH from NJ

C:\>tracert ovh.com

Tracing route to ovh.com [213.186.33.34]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 192.168.1.1
2 7 ms 7 ms 7 ms 10.63.192.1
3 6 ms 7 ms 6 ms dstswr2-vlan4.rh.brfdnj.cv.net [67.83.242.194]
4 * * * Request timed out.
5 13 ms 13 ms 19 ms 64.15.2.97
6 9 ms 11 ms 9 ms 64.15.0.145
7 * * * Request timed out.
8 * * * Request timed out.
9 * * * Request timed out.
10 * * * Request timed out.
11 * * * Request timed out.
12 * * * Request timed out.
13 * * * Request timed out.
14 * * * Request timed out.
15 * * * Request timed out.
16 * * * Request timed out.
17 * * * Request timed out.
18 * * * Request timed out.
19 * * * Request timed out.
20 * * * Request timed out.
21 * * * Request timed out.
22 * * * Request timed out.
23 * * * Request timed out.
24 * * * Request timed out.
25 * * * Request timed out.
26 * * * Request timed out.
27 * * * Request timed out.
28 * * * Request timed out.
29 * * * Request timed out.
30 * * * Request timed out.

Trace complete.

Excellent - did I ask you to ping them?! You won't be able to cos they put their shit on like super stealth mode. (edit btw, ovh.com is delivered by 50 dedicated HG servers - their own admission. It's never going offline. But it does also serve the entire Manager control panel that lets clients control their servers. How they do that super stealth mode though is beyond me! But it is required because the client control of their servers depends on it.)

No, I asked for download speeds from those two URLs :winkwink:

In any case, here's a reverse from ovh to you (well, your last public IP)

Looks good to me.

:1orglaugh:1orglaugh:1orglaugh
100%

meanwhile, marge knows what it's all about. You can see her controlling her asterisk phone from the driver's seat....

i just wanted to see EU server in 5-6 hops in traceroute from US like you said earlier. my trace shows that on 6th i'm still in US and lost somewhere in transatlantic pipes afterwards. i'm not far from NY pipes. that's all.

trace borkedcoder.com then

but that direction is not important in a US client base setup, it's OVH to US.... that is the path that is important for speeds.

--edit, I'll send the reverse traceroute if you do it in the next 5 mins....

i've seen worst in US heh, although i'm not sure how to treat long Request timed outs at the end of traces, yours:

C:\>tracert borkedcoder.com

Tracing route to borkedcoder.com [178.33.218.112]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 192.168.1.1
2 7 ms 9 ms 8 ms 10.63.192.1
3 8 ms 10 ms 7 ms dstswr2-vlan2.rh.brfdnj.cv.net [67.83.242.162]
4 * * * Request timed out.
5 9 ms 9 ms 9 ms 64.15.2.125
6 161 ms 101 ms 198 ms 64.15.0.65
7 14 ms 14 ms * paix.ny.routers.ovh.net [198.32.118.106]
8 161 ms * 260 ms ldn-1-6k.uk.eu [213.251.128.30]
9 91 ms 85 ms 92 ms rbx-g2-a9.fr.eu [91.121.131.177]
10 86 ms 104 ms 85 ms vss-3-6k.fr.eu [94.23.122.93]
11 * * * Request timed out.
...
Trace complete.


back to SSDs, i'm looking forward trying them out on my workstation later this year

The first link I hit 700KB/s

The second link I hit 1.1mb/s

This is on a comcast 16mb/s cable connection in Minneapolis, MN USA.


Also, I've ran my own servers since about 1997 and I've never used managed servers. Always unmanaged. So your 99% guess is probably a little off. Maybe not by much, it seems like everyone on GFY uses managed hosting which really surprised me when I joined here.

Not many people use *DSL in the states anymore, let alone anything higher than 3-5Mbps our copper infrastructure is horrible.

Take it how you see fit, but thats no where close to what I said. Arguing with people on GFY is pretty much pointless because this is going to go round in circles.

Now quick, everyone jump ship and move their stuff to OVH. Quick! :disgust

Make sure you research the SSD drives you are looking to get, look for something with 280+MB/s read speeds and 250MB+ write speeds. I.e OCZ Vertex/Vertex 2 and Turbo or Intel X series SSD drives.

Of course you won't say those words, but you admitted they have great peering with ISPS:

Ergo, a well-chosen CDN company, combined with OVH will delivery outstanding results. If you don't think that, please explain why?

The first server is in LA (don't know why I kept saying Sand Diego before!), the second is in Paris. So, OVH transit to Comcast works well. :thumbsup

and to think the copper wire came out of the US thanks to Doolittle (what an oxymoronic name!)