Yeah. For years I was happy with DigitalOcean and had no issues, but this is the interaction I'm currently in the middle of:
I can't reach some particular other server; both of us can reach everyone else on the internet, just not each other
Contact support sending traceroutes
Support asks me to send mtr output (I mean... it's a little more detailed than the traceroute I guess, sure what the hell)
I do
Support asks if I'm running a VPN on my Linux server
Support tells me to disable my VPN
Support asks me to provide mtr output from a third place on the internet (which... he could do himself... I consider telling him to do it himself but decide not to). I send the mtr output
Silence for 2 days
I mean, it's far from an emergency but it doesn't inspire a ton of confidence. 🙂
Idk man namecheap has been doing it for a LONG time. 24 years.
Not clear on Vps probs because they want to know if you’re trying to run a home server, or a web hosting business where you’re going to house 47 clients, or if you’re trying to run a fucking public education system or something.
Name cheap ain’t obscure. Two cows was obscure. They were fucking awesome. But again I have no current Intel on them.
Edit: I don’t know man, I just checked, the pricing seems pretty clear to me
Maybe I’m just out of the loop. The pricing page seems clear but I still don’t trust it. Additionally I don’t care if the server is running on Raid 10 or not.
Linode has been ok for me. I wouldn’t use their managed services. If you need that type of scale go to AWS.
One cool thing I experienced with linode yesterday is I could boot from one of their kernels instead of my own grub kernel. I had a bad update on alma9 and this made recovery much easier without having to rebuild the node.
I was very happy with Vultr, but they weren’t happy with my Linux isos.
Using linode now, the backup system is more expensive and the emulated terminals don’t seem as nice as I remember vultr but I’m still very happy with it, especially for the price.
Got it, thanks for the recommendation, I poked at both of them a little and it seems like what I'm looking for.
Also, out of curiosity I looked at Vultr's GPU options and said "oh sweet they offer options with tons of GPU RAM if I want" and then looked over and saw it was $3,500-$14,000 a month.
Okay, I just signed up and from the tiny bit I see I really like Vultr.
Also, for new customers they have discount codes. I went with $250 credit for new customers, and then after I signed up, learned that it only applies to money you spend during your first 30 days.
LOOKS LIKE MACHINE LEARNING'S BACK ON THE MENU BOYS
Edit: Also, this is the kind of thing that tends to inspire confidence:
I’d recommend against ovh. We host lemmy.ca there and that’s been fine so far, but I had a friend whose production server was down for 3 days due to a switch problem.
I used to have that issue, my screen would flash 6 times and display a bunch of symbols like those. Gave me a burning headache and next thing I know, I’m waking up in an alley of an unfamiliar city halfway across the globe with somebody else’s blood coating my hands and clothes.
I feel like Broadcom is aiming for cloud-like pricing for on prem services with none of the other benefits inherent to an Azure or AWS deployment. Not exactly the way to hold onto clients.
I’m familiar with proxmox and the broader KVM ecosystem. I’m also a huge fan of Veeam, who have said they’re exploring support for proxmox. Shouldn’t be too difficult to implement, given they have a RHEL backup product already.
I think Broadcom intends to dig VMware out of dept to turn it into a profitable company. This means killing off the smaller customers as 90% of the business comes from enterprises that will never switch to anything else no matter the cost.
It’s really difficult to move away from a backup software you just switched to and paid > 100k to license for the next 3 years from a leadership standpoint haha. PBS, zfs snapshots and send, Ceph duplication. It all does more or less the same thing.
Proxmox is missing a lot of enterprise features. If you run a virtualized data center, it’s really not going to cut it. OTOH, if you are a small operation with just a handful of virtual servers, it might be “good enough”.
The obvious alternative was Hyper-V, but it looks like MS is already killing it to force people into Azure.
When you look at enterprise-level hypervisors, there really aren’t a lot of options.
The two big ones I see is no official vGPU support (you can get it to work unofficially but it’s not prod ready) and the clustering scheduler is still in active development while still missing several features that vSphere’s scheduler offers.
I’ll tack on just a bit from here, and maybe someone can correct me if I am wrong.
VMware’s HCI clustering is far better than proxmox + ceph/other.
VMware’s NSX network virtualization enables their fancy HCX site orchestration.
Even without NSX/HCX, Site Recovery Manager makes for a slick redundancy/fail over option.
VMware’s EUC option, Horizon, beats the absolute pants off of Citrix. And that was Citrix’s whole game.
The vGPU option first lived in EUC, but turns out scalable GPU sharing is just plain useful.
And then there is the orchestration management, allowing for power savings, automatic balancing, and more.
Basically, every high level solution they had on their platform was without a true parallel, and was built on a rock solid foundation. Even if their support is shit(it is), the platform is so ubiquitous and approachable that you could just use their support as an insurance of sorts, and it gave upgrade rights through the years.
Broadcom knows who uses those high level features, and knows they’re stuck. Our options are a full cloud migration, loss of features, or pay up. They’ll disregard every customer small enough to not need any of that, and they will milk every customer that’s too big to go anywhere else.
If you’re one of the small folks, I’d say look into proxmox, openstack, xcp-ng, or have a path to cloud in mind. If you’re one of the big folks, I recommend Balvenie, Macallan, or Johnnie Walker, cause you might as well enjoy a good drink if you’re gonna get fucked.
Sysadmin
Active