He mentions benchmarks but provides no statistics. The pageview times could be heavily influenced by few users, say the author repeatedly accessing his own site. There is nothing concrete enough here to make me consider moving.
Interface is a bit dated but completely functional and not that bad to deal with. I do believe they are building a new Manager, but I feel like they've been working on it since before the DDoS attacks late last year. Not sure when we will see it or if it is even in development anymore
Re the hack. It was a while back, but was troubling. However, they're willing to sign HIPAA Business Associate Agreements, which for my company's purpose mitigates those concerns... they're willing to be on the hook for host & physical security.
Re the DDoS. As a mitigating factor, the DDoS was rather intense. It actually took their upstream provider offline. They supposedly have mitigated it with redundant connections and orders of magnitude more bandwidth at the two hardest-hit locations (Atlanta & Newark).
That said, time will tell. I love pretty much everything about their service (hardware, interface, bandwidth, cost), but I've got a business to run... if there's more extended downtime, I'm gonna have to move, though I won't like it.
Digital Ocean and Vultr were the two I was looking at, though don't really know much about the popular opinion of Vultr.
A hack is "troubling", and you deal in healthcare? I work in med-tech. If my hosting service was hacked and PHI potentially exposed, we'd have finished migrating away within a week. If your concern is only a business agreement to limit liability, do you mind sharing your product so I can never use it?
Luckily, we don't deal with patient records or related PHI; which lowered concerns to the "BAA is sufficient" level.
If I was running an EMR or such, I'd want the servers in a colo, with gated physical access, and contracts signed in blood of the each person's firstborn :)
Interesting comparison. It's been a while since I did a shootout but I was really surprised and the huge differences between the different providers. We run a little Ruby thing and we tried benchmarks on like 10 different places 2 years ago and found Softlayer was the best performance for what we do for the money. That is to say we could run X apps on an X sized server that costs $X per month at Softlayer while it costs $X+$N everywhere else to get the same performance.
The only surprise here is i couldn't immediately find their Linode or Digital Ocean referral link.
> Of course, these stats should be taken with a pinch of salt, as the old VPS was located on the East coast of the States, and the new one is located on the West coast.
oh hey, that 4x performance gain? i think we found it.
I have VM's at Linode, DigitalOcean and RamNode. My VM's are long running (vs. bursty spin-up, do a job then terminate)
I prefer the UI and feature-set of Linode. I have found the technical staff of Linode to be superior to others.
I find the CPU and disk performance of Linode and DO to be about the same.
RamNode is cheaper with similar CPU performance, but their UI is qwerky and I have run into billing problems with them. Their support are always fast to fix, but they shouldn't have to in the first place.
Wasn't it just last year during the holidays nearly all Linode hosts were offline, for like days? I'm a DO customer and there is absolutely no way I would consider switching to Linode.
It's a bit strange that the VPS would have something to do with it. What I mean is that Digitalocean is basically sitting side by side with google. I ping'ed google from the NY3 and I get really fast response time.
booteNg1iethahb8uh
Oops that was my password that I pasted... here is the good paste.
root@debian-512mb-nyc3-01:~# ping www.google.com
PING www.google.com (172.217.4.36) 56(84) bytes of data.
64 bytes from lga15s46-in-f4.1e100.net (172.217.4.36): icmp_seq=1 ttl=59 time=2.36 ms
64 bytes from lga15s46-in-f4.1e100.net (172.217.4.36): icmp_seq=2 ttl=59 time=2.44 ms
For me, a couple things I'd like to see more of from some of these smaller VPS providers... block data storage that can be mounted to the filesystem, offering redundancy and backup space to target. DO now has this as an option... outside the big players (AWS, Azure, GC) I haven't really seen it.
Beyond this, is managed SQL option... I mean, I'd prefer PostgreSQL + plv8 at a minimum, but open to others. I really don't want to manage this.
The fact is, I'm not willing to dedicate hundreds a month to hobby and one-off projects, so continue to DIY, but would really like to see more options out there... maturing docker management is pretty compelling as well. It feels like AWS, GC and Azure aren't really even trying to compete, because the additional services they offer have no real competition.
31 comments
[ 3.4 ms ] story [ 76.5 ms ] threadI remember Linode getting hacked a few years back. Also, how is the user interface on Linode?
I just don't switch providers overnight to save $20.
A. Linode has gotten hacked atleast once before
B. They were suffering from those constant DDoS attacks for a week or two a while back
Those two, and their resulting responses to those events have permanently made me have reservations about using their services.
Re the DDoS. As a mitigating factor, the DDoS was rather intense. It actually took their upstream provider offline. They supposedly have mitigated it with redundant connections and orders of magnitude more bandwidth at the two hardest-hit locations (Atlanta & Newark).
That said, time will tell. I love pretty much everything about their service (hardware, interface, bandwidth, cost), but I've got a business to run... if there's more extended downtime, I'm gonna have to move, though I won't like it.
Digital Ocean and Vultr were the two I was looking at, though don't really know much about the popular opinion of Vultr.
http://arstechnica.com/business/2012/03/bitcoins-worth-22800...
https://blog.linode.com/2013/04/16/security-incident-update/
https://blog.linode.com/2014/01/19/an-old-system-and-a-swat-...
https://blog.linode.com/2016/01/05/security-notification-and...
I'm not even sure if they ever figured out the last one.
So yeah, hopefully ryebit was just kidding.
If I was running an EMR or such, I'd want the servers in a colo, with gated physical access, and contracts signed in blood of the each person's firstborn :)
> Of course, these stats should be taken with a pinch of salt, as the old VPS was located on the East coast of the States, and the new one is located on the West coast.
oh hey, that 4x performance gain? i think we found it.
I prefer the UI and feature-set of Linode. I have found the technical staff of Linode to be superior to others.
I find the CPU and disk performance of Linode and DO to be about the same.
RamNode is cheaper with similar CPU performance, but their UI is qwerky and I have run into billing problems with them. Their support are always fast to fix, but they shouldn't have to in the first place.
> Ubuntu 14.04 on the first one, while it is a fully updated 16.04 now
Oh yes, what a completely fair and unbiased measure of response-time improvements, much like extolling the virtues of moving from apples to oranges.
Publicly confirmed Linode hacks: 4 at least, probably more
Times Linode got caught trying to cover up the hacks: At least 2 (The HTP one and their most recent publicly acknowledged compromise)
I will never use DO because I haven't forgotten about:
1) Their incredibly rude response to "we don't zero drives by default" on GitHub, effectively blaming the customer with "it's a feature."
2) Their censorship of a blog by an employee, at request of that employee's friend.
https://news.ycombinator.com/item?id=7016735
booteNg1iethahb8uh
Oops that was my password that I pasted... here is the good paste.
root@debian-512mb-nyc3-01:~# ping www.google.com PING www.google.com (172.217.4.36) 56(84) bytes of data. 64 bytes from lga15s46-in-f4.1e100.net (172.217.4.36): icmp_seq=1 ttl=59 time=2.36 ms
64 bytes from lga15s46-in-f4.1e100.net (172.217.4.36): icmp_seq=2 ttl=59 time=2.44 ms
Beyond this, is managed SQL option... I mean, I'd prefer PostgreSQL + plv8 at a minimum, but open to others. I really don't want to manage this.
The fact is, I'm not willing to dedicate hundreds a month to hobby and one-off projects, so continue to DIY, but would really like to see more options out there... maturing docker management is pretty compelling as well. It feels like AWS, GC and Azure aren't really even trying to compete, because the additional services they offer have no real competition.