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.
Amazon Lightsail 1GB is no match for $10 VPS from Linode, DO, Vultr... etc.
voldermort
Member
in General
Disclaimer I am in no ways a fan of any services, the article is opinion of author of article and I have no connections to author at all.
I wonder if someone here has experienced Lightsail and willing to do benchmarks with some famous VPS providers like Kimsufi, OVH etc.
Comments
But you don't understand! It doesn't have to actually work, what's important is that it's Enterprise.
I don't need my web server responding quicker than in several seconds, it just needs to have the certifications and compliance.
You kids don't understand there's a whole world outside LET. And nobody out there needs more than 10% CPU usage anyways.
It's not like you should expect any hidden limits to be declared upfront, what matters is that with Amazon you get the entire integrated infrastructure built from ground up.
Did I cover all of the Amazon fans' bullshit excuses yet?
Goodluck to use 1 CPU for webhosting as at some point it will fail, try to host the wordpress.
There's usually a little API dick-length measuring.
The number of cores alone doesn't tell the whole story. You only get one core at Digital Ocean, Linode, and Vultr too on their $10 plans but plenty of people host relatively high-traffic sites on those without a problem. The OP's link even says a (one core) $3.49 OVH VPS was able to handle the load just fine.
Hell, even on this 3/year NAT vps you can use 15% cpu: https://www.wishosting.com/plugin/support_manager/knowledgebase/view/11/terms-of-service/
How to get root pass instead keypair? Just useless
@jenkki maybe just type passwd at the prompt and add one
You're acting like there's no use case scenario for AWS, and that's a bit foolish. I have a handful of things on AWS, because they sit there mostly idle and I just want them to be up 100% of the time (like IRC services, or redis).
If you're running CPU intensive things (I tried running sensu-server on AWS t2.nano.....it was a nightmare) then its not a good fit. But hell if you wanted to run a seedbox or some massive website then DO/Linode/VULTR might not be a good fit either.
You can buy AWS credits at 10% of their value, so a 512MB ram, 10%of 1 core, AWS VPS for $6/yr? Hell yea.
Understand that not everyone's needs are the same as your needs.
EDIT: Here is my redis server on AWS: http://i.imgur.com/QJ9w65e.png Tell me why that needs 100% CPU.
Event coupons?
Yea. http://i.imgur.com/SX8V4QN.png
EDIT: For laughs, here is my sensu-server on AWS t2.nano: http://i.imgur.com/gqUL1Up.png
All of that red activity is CPU STEAL, just looks weird on the graph.
Who? What? Where? When? How? Link plz.
Is wishosting reliable?
Technical perspective is not that powerful in decision making. Boss will choose household brand name. And Amazon is always trusted by non technical managers.
And as a subordinate, you don't want to make an opposing suggestion, because it will be your butt on the line. It is better to have the manager's butt at stake, rather than yours.
And Amazon has more than 3 people working for them when problems arise. For some people low cpu usage is fine for development, prototypes, and initial implementation along with high revenue low use cases. Plus its a one stop shop to get CDN and latency based DNS along with RDS to easily get redundant Database by clicking a few setup features. So lightsail does have a use case and everyone seems to overlook this is their first site and first entry into the marketplace for all inclusive VPS. Do you honestly think Amazon is going to remain stagnant? At anytime they can turn the AWS Staffing and Amazon Financial muscle on the marketplace, whenever they deem appropriate. Next years offerings will undoubtedly be better than this years.
You will get blamed anyway, when the server you are supposed to be operating is too slow.
It's a fair point that blaming the host is easier when you didn't personally vouch for them and your boss chose them. If you're in that kind of business relationship where your boss wants blood every time something is less than perfect, perhaps anything that eases the transfer of blame has value to you.
Perhaps, though, that person wants to keep their resume on hand and updated at all times.
You also didn't investigate into what performance figures were to be expected with that offer, and with the benchmarks that are widely available which show things like 60 MB/sec IO and 2 MB/sec network, and now with this CPU throttling scandal, failed to warn your higher-ups about any of that, demonstrating gross negligence of your job.
My bosses are in bed with Microsoft and it's Azure all the way for us but while it got problems it isn't as bad as Amazon & Google.
Their support was non existant compared to what Microsoft offered us.
This is very true.
He wasn't asked to. In the vast majority of small businesses, where to host isn't a decision for which someone gets to investigate with benchmarks and stuff. "Where do we host?" "In the cloud -> Amazon".
In an ideal world, yes. In practice, "nobody ever got fired for choosing IBM" applies to Amazon too.
Yes, that. A crappy underperforming product people will buy just with a single goal to cover their own ass and not "get fired" -- an excellent image for Amazon. And never mind that the setup they will get of course won't perform anywhere near decent, and their business as a whole will end up losing customers due to providing a poor user experience. After all it's not like anyone has even considered to pause and think or compare, they just went with Amazon because that's what everyone does.
Right, but I feel like you're missing the distinction between "negligent and responsible for fucking up", and "being held accountable for fucking up". While the former is definitely true if you blindly go with Amazon, the latter really isn't in most cases.
All ethics and business consequences aside, picking Amazon is a great idea from a job security perspective, because you're never going to be held accountable for that choice, even if it turns out to be a poor one.
If I had a penny for every time I heard somebody say "Amazon can't be that bad, otherwise why would everybody be using them?", I would probably never have to work a day in my life again.
One benefit of AWS lightsail is you don't need to worry about them going out of business or being acquired. I ran into that problem to many times on what I thought were solid LEB hosts, and I have decided to permanently move away from the smaller LEB hosts. I decided to go with Vultr (hopefully they will stay around for a good amount of time).
If I could I would highly consider lightsail, except right now their services are not yet as good as the medium sized hosts(Vultr, Linode, DO). Lightsail does not have CentOS, has slow IO, has cpu throttling, etc.
I've been on that other side too, where I convinced people to let me go with less popular but more solid options on things. The win on money spent and service quality did not outweigh constantly being blamed for my "poor choice" every time anything went wrong. Every time... "_____ called me and said they filled out the contact form. It didn't work. Is this because you went with _____ instead of _____."
Idiots are idiots, but life is short, you pick your battles. Some hills are not worth dying on.
Exactly what I am usually concerned about. But I think Kimsufi and OVH are really good cheap alternatives for same price point. People here might add more.
Then just don't bother with these "one guy with a server" operations. They're up not against AnthonySmith, KuJoe or NickA, their competitors are OVH, Linode, DO, even Vultr. And those aren't going away anywhere.
Amazon is a big, brand name - but that doesn't mean they won't spin off or even kill their new VPS line. Big companies kill their lower margin product lines all the time.
And look at the Soviet Union: once a global superpower of nearly 300 million people. Now poof! All because the price of one single commodity crashed in 1985/6
I dunno if I'd call it that: the cpu throttling seems to be the same thing they've been doing with the smaller ec2 instances for a long time. They have a page describing exactly how it works ("CPU credits"). I've done ok using their high-$$$$$ instances for heavy computation but the smaller ones have never been that great.
Meh. There's a lot more to Amazon than just the compute they offer. The supporting products win hands down every single time. Plus for performance- it's provisioned past the t2 series, so if you want insane performance, then configure and pick the right instance.
I did use those big instances because I worked at an AWS-using shop so that's what we did everything with. I had to do a computational task for a week or so and the instance we used (don't remember exactly what it was, but comparable to a mid-to-high range LET dedi) cost somewhere north of $1 an hour, i.e. $700+ a month. Similar compute power at Hetzner is more like $30 a month. At least it wasn't my money that was being spent.