Hacker News Reader

Comment by taligent | Hacker News Reader

Err. Linode has pretty average performance compared to the SSD providers.

And during my time as a customer Linode was the worst out of all the VPS I had ever tried up until that point. Constant outages at Fremont and heard about the same at London DC as well.

View on HN

You mean the SSD providers who cost many times more than Linode? I've tried a few SSD-based instances. Storm is pretty good, but considerably more expensive. Cleverkite and Digital Ocean are both far worse. I wouldn't trust any valuable infrastructure to a provider who offers fast I/O on a crippled network, or who only lets you do fast I/O for a couple of seconds before your instance gets throttled into oblivion, even if they were free.

I see that you're a DO fan, but if you've actually used them and aren't just promoting them out of undisclosed self-interest then I'll bet you even know which of those two descriptions fits them. Being buzzword-compliant doesn't make them a good provider.

I switched from Linode to Digital Ocean due to price, but I'm not familiar with which of those two descriptions fits them. Could you elaborate? I haven't experienced anything yet that would lead me to believe they are "far worse;" their network doesn't seem crippled, and my IO-heavy site (imageboard) isn't throttled.

Wonder how sustainable Digital Ocean's pricing is in the longer term.

I'm running a VPS with RAMnode and it's been nothing but roses.

Here's a benchmark on my VPS http://serverbear.com/benchmark/2012/11/20/UMLXH3qNTNOyfbGF

And here's a bunch of SSD VPS plans sorted by IOPS http://serverbear.com/compare?Sort=IOPS+Seek+Benchmark&O...

That's a really handy list. Thanks!

I did give RamNode a try today after seeing them mentioned here, and they do indeed seem to do a good job of delivering on the SSD promise. I'll probably stick with Linode in NJ instead of RamNode in Atlanta (because the way I use my machine means that network latency to places near my home matters a lot more than disk performance) but it's nice to know that at least one vendor finally seems to be getting it right.

Thanks for this link. Thats a very helpful site.

I am choosing between RAMnode and Digital Ocean. Both look pretty much affordable and recommended.

Any tips?

Just logged in to one of my servers in the London DC. 952 days of uptime, so in my case it has been pretty stable over the past couple years.

That doesn't take network outages into account, which they may or may not had.

I had around 600 on one of my machines in Newark until I forgot to update my credit card info and they shut them off.

Same here, I've never had any outage issues with the London datacenter.

Though I had one issue where I couldn't install 64bit image to the server. I sent an support ticket about it and in less than ten minutes the whole thing was sorted out.

You prompted me to check mine. 994 days of uptime.

No downtime at all since I got started with them in late 2007, except for a few reboots to activate free loyalty upgrades and one paid upgrade. I've never lodged a support ticket that was left unanswered for more than a minute or two. Very predictable performance and reliable support are huge.

How are you able to achieve such long durations of uptime ? Don't you need to reboot your machine for updates ?

My attack surface on that particular machine is sufficiently low (everything off except keyed SSH and nginx) and the machine's duties so minor that I don't bother.

Reply on news.ycombinator.com