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
The dino should be wearing a striped sailor shirt.
i paid for a ks-le-e hdd version,, but got standard spec server delivered,, it seems ks-le-e dont have much upgrade this time ,,
Yeah seems like the LE-E are all as advertised.
Where?
Gra
On one of ks le e , ssd version , i got prety new disks, 1500 hrs
That's decent. My BHS was delivered with 10 year old HDDs and RBX with 4 year old HDDs. No guesses needed to know which I'm gonna refund.
I know finding disks with high hours seems like a red flag.
But these disks are usually enterprise grade and still hold the fort.
One can always request replacement disks via ticket or api. But there is always a possibility to end up with even more wearing
My disks from old server are over 80k hours but no issues and chugging along.
If you need new disks then maybe ovh is not the right provider to gamble on.
Still trying to find a KS-LE-B NVME lottery edition to serve as a home for my rare media collection.
Please do not give up, your order is important to us…
You're right. Will go shopping later and see if I can find a nice shirt.
And I even have an SSD in a server that has more hours on it than could be possible (about 120k) - the ssd model itself got introduced a year (2013) later than today minus those 5000 days. Maybe the provider got hold of a preview model?
Still running
Also have HDDs with 80k plus running fine while other (Enterprise) HDDs with ~20k died.
Interestingly, the situation from Finland is the same for all OVH EU locations (FRA/LIM/GRA/RBX/SBG/WAW), route goes through FRA (fra-fr5). The additional hop doesn't mean all that much though, my ping is 33 to FRA, 34 to LIM, 35 to SBG/WAW and 39 to GRA/RBX. I have Telia as my residential connection which is a Tier1 ISP so I'm tapped directly to Tier1 peering (Telia Carrier/Arelion/Twelve99/AS1299) and the first hop in traceroute that is outside Telia's own network is from AS1299 (Telia) to AS16276 (OVH) in Frankfurt. The full route is HEL > STO > FRA > (LIM/GRA/RBX/SBG/WAW).
With such excellent connectivity however there is a bewildering issue, while I get full speed to all OVH locations through the iperf speedtest at proof.ovh.net, the upload speed of my KS-LE-B (last year's with 300M uplink) in GRA to my home is capped at exactly 50M, while the download speed is uncapped. It isn't an issue with this specific server either as the exact same thing was true with my earlier KS-1 server in GRA bought 10+ years ago and only sunsetted this year (the very first KS-1 offering, Atom N2800, 2G RAM, 100/100 link). I never paid that much mind to it until now but it is interesting.
And yes the YABS tests get the correct results for its testing locations, does 1000/300 for the KS-LE-B and did 100/100 for the KS-1.
Note that this is quite uncommon for me as usually I get full connection speed to pretty much anywhere in the world and back-in-the-day before I even knew what BGP/routing/peering/tiers even was I didn't even consider that links between carriers could be throttled or routing otherwise affect your link speed and always assumed that whatever speed I'm getting if it's not maxed out its solely based on the connection speed at the end destination.
And even more interestingly, when I put my server behind CloudFlare this issue is entirely gone. Note that I have tested this correctly as in using a new resource so it doesn't just retrieve it from the cache, so a brand-spanking-new file on the server behind CF gets full upload speed from the server to the local CF node that is inside Telia (AS1299) and finally to my PC.
This means that for some reason, even though my connection to OVH AS is as direct as it possibly can be since my own Telia AS routes directly to it, the upload connection from OVH to Telia for the Kimsufi servers (but not for the proof iperf server) gets throttled at some point, while whatever route CF uses to get the data from OVH AS to its node within Telia AS here, does not. This could very well be due to some peering contracts considering Telia is a Tier1 network, like any other Tier1 net its peering policy isn't completely open and maybe the peering contract between Telia and OVH isn't open?
And in fact Telia did sell its peering network to a Swedish infrastructure company in 2021 and that's how it got its new name "Arelion". What worries me here that of course now the new peering company expects a return for its' investment and looks to extract profits from the infrastructure itself, which is a step away from the original model how Internet developed in Europe and North America that is first there was university networks freely peering amongst each other, then commercial ISPs stepped in offering plans to customers for internet connectivity but most importantly the idea of building your network and then freely peering with other carriers stayed, peering itself was operated essentially as non-profit and the profit motive here was that the ISPs of the world get to make profit from their end-user customers, while the peering between carriers stays free to keep everyone's costs down. This is the reason why Asia, South America, Africa and Oceania have so fucked up transit costs because they came into the game later and realized that there is money to be made in peering and now in addition of ISPs charging for internet access the greedy peering companies charge massive transit costs for the peering itself.
There are these two excellent articles about worldwide transit costs here in CloudFlare's blog
The Relative Cost of Bandwidth Around the World
Bandwidth Costs Around the World
A good read for anyone that is interested in how worldwide Internet works and what is ultimately responsible for things like why servers cost more in Asia than in Europe.
So what I'm worried about is that there are vultures swarming around trying to make it so they could extract profits from peering and end the days of free transit that makes connectivity fast and cheap in Europe and North America.
In fact this is what the fight about "net neutrality" boils down to too, and what the corrupt former FCC chairman Ajit Pai tried to do and destroy net neutrality, import the "Asia model" to the U.S. He might sincerely have believed to be doing nothing wrong though, considering that corruption/kickbacks is how virtually all business and politics are done in his home country of India and the concept of common good doesn't exist in the culture.
In any case though, considering that the LET community is global, making a thread where people could post their worldwide traceroutes, pings and speeds to various OVH locations would be interesting.
Holy crap, that was a long post. Hopefully at least someone found it interesting.
At least some KS-LE-B servers are equipped with relatively new drives, which is good thing, as OVH’s eco series typically uses hardware that has been refurbished or recycled multiple times.
Perhaps Octave needs to approve ever order manually. Let's Tweet him, I'm sure he will appreciate our support.
@anzz1
As I read about 75% of the post:
I'm on the opposite: throttled as hell on CF; fine on OVH (but yes, I'm in Germany, so OVH is fine and yes, I use DTAG and they don't really peer with CF).
got 600, was pretty surprised.
Rest NVME has like 30-40k hours.
I only asked OVH once about an NVMe replacement as one had died and the other one was in pretty bad condition: got two brand new disks as replacement.
Could be luck (while brand new is sometimes also risky).
But I only ask for replacements if real errors (like reallocated sectors on HDD) turn up. Age is not that important with DC/Enterprise disks (may be on consumer ssd or nvme).
WAW DC.
until OVH figures, they can't deliver.
Tag me in 1 year.
I've got a last invoice in RBX, I might pay it, I absolutely love it when OVH sends me an email to apologize (except when KS-A is involved)
The way that I cannot keep up with @Saragoldfarb ’s Dino’s fashion changes 🤦🏼♂️
Want to be three days late again?
We should tag you in 3620* days
*) two days earlier to prepare.
On one of my servers I got a pair of NVMEs with only like 4TB written
its very rare,, usualy nvmes are around 40k+ written,,
I missed the KS-LE-E. how likely is the restock? (looking for one with SSD in bhs)
service-transfers category is lit over the past few days.
FINALLY DELIVERED RBX ON CA ACCOUNT
COMPLAINING WORKS CONFIRMED
Wooooooot
I never got a chance to get rbx
Wanna transfer it to me?