Amazon EFS...NFS storage on AWS
- Thread starter thedude
- Start date
MattW
Well-known member
Because I also use OVH, and they are 100% un-managed. They shouldn't log into your server at all or need the root password.
Moscato
Active member
Deeply discounted AWS EC2 is still more expensive than the standard price at linode or digital ocean
Moscato
Active member
It seems like the only situation that AWS wins is:
If you don't care about the money
You don't want to use a security CDN like cloudflare or incapsula
You have lots of storage needed
You aren't considered about outward data consumption
You want auto scaling that doesn't require scripting
If you don't care about the money
You don't want to use a security CDN like cloudflare or incapsula
You have lots of storage needed
You aren't considered about outward data consumption
You want auto scaling that doesn't require scripting
EQnoble
Well-known member
Nope, since both were VPS and even though the first was supposedly not mine anymore, it remained attached to my account but all my access to it was removed and I was held responsible for it.
Also I changed the root password for the 2nd one from cli rather than from their control panel and then contacted them and I was told that I need to change my password from their customer dash because 'they can't log in now'.
To further that, when they sent the email saying I was banned and all communication would be blocked from me, they didn't say 'we closed your server'...they said 'we closed all of your servers under your account' which doesn't even make sense. When I was left with no other choice to have an investigation opened against them and told the credit company that they had taken my money, not offered a refund and blocked all forms of communication from me in response me reporting suspicious activity...they had to ask me to repeat if 4 times because they couldn't believe it either.
A question, why are their US server offerings actually in Canada eh...US is not the same NA and should be labeled appropriately...to me it just means they want to attack Americans and be shielded from our laws.
Rather than doing the responsible thing and firing that employee, they went code red and united against a paying customer to cover internal criminal activity within their company.
Again I am 100% documented with my actual experience with them and the screenshots with video and phone recordings were enough to get ALL my money back from them (even though my credit company upfront said I probably cant get my money back because it is a service) because after my CC company contacted them they found that they wouldn't respond to them on my behalf either so...yeah OVH can get a big rubber one in their stocking this and every Christmas.
MattW
Well-known member
Ah, that's different then. I only use their dedicated hardware, which is 100% unmanaged. That's the only reason I was asking. Nothing untowards, or questioning your experience with them.
Jim Boy
Well-known member
In other words if all you want is a simple server or two, Linode is ideal. Anything more and the work 'hack' becomes useful.
EQnoble
Well-known member
No doubt.
In all fairness, I do manage (dedi) servers through them for 2 different people (when they need me) and have since before I got into programming in general, but at the time I was communicating in French and I dealt directly with OVH out of Europe and not a subsidiary/child/offshoot of the company closer to my landmass.
The reason I had went with a VPS with them here for my own needs is I found that they finally offered something for us in the USA and my experience with them and services offered overseas up until the unfortunate event were excellent and the people I managed these servers for REALLY hammered them and the pipes they were on and as long as the bill is paid and they don't do dirt all is good in the hood.
The only reason I didn't move the two parties off their services is that they are customers of many years and are not the ones with a bullseye painted on their back (they don't share my legal name and don't live in the same state as me) and the reality is I know that a single bad apple that messed with me and the way they handled it implies the reality that I personally can never trust them with services in my name all things considered.
My warning against them mostly pertain to their "American" offerings as their bread and butter are the EU and Asian markets and they are simply better organized than the offshoot running in Canada and are not foolish enough to do what their Canadian counterparts are doing. My suggestion to them is if they want to actually take the US market seriously that they need to send a handful of experienced OVH staff and a bilingual exec (French + English) with proven history within their company to oversee the Canadian operations so that it is not running like a wild west high-noon showdown where a customer needs to walk around with their digital six shooter to protect themselves from the provider.
That said I am now using Softlayer (IBM) and I can say that I am completely happy with their services and their API is pretty damn diesel and it allows you to do everything you can do in the dashboard and more. From the SL dash I can add whole servers, change up resources, grab IPs, etc... which makes for a do it yourself type of persons ideal situation. I add 8 gigs of memory and have it within minutes...that is service. I have even gotten phone calls from them literally asking me if everything is going ok, if I understand the abilities and implementation of the API and wondering if there is anything I think they could do to improve my experience. I have a direct contact within their company and I have been invited to contact him with any questions, issues and suggestions that I have which might be better suited in the form of speech over text based tickets.
WSWD
Well-known member
The automatic scaling is great if you have a small site that really isn't going to do anything or get any visitors on a regular basis.
You pull 10TB of outgoing transfer on S3, and you're looking at a $900 bill...for one month, just for the bandwidth!!! You could get 3 very beefy dedicated servers for that price, and never have to worry about scaling, because you would have more server power available than you knew what to do with. You could power some of the larger sites on the Internet with that setup.
This is where places like Amazon (and Softlayer) just boggle my mind. Bandwidth has gotten dirt cheap here in the US, yet they keep charging ludicrous prices for it. Softlayer's prices on bandwidth have actually increased! WTF?
So S3 for any decent site makes little to no sense to me, unless you just have money to burn.
You pull 10TB of outgoing transfer on S3, and you're looking at a $900 bill...for one month, just for the bandwidth!!! You could get 3 very beefy dedicated servers for that price, and never have to worry about scaling, because you would have more server power available than you knew what to do with. You could power some of the larger sites on the Internet with that setup.
This is where places like Amazon (and Softlayer) just boggle my mind. Bandwidth has gotten dirt cheap here in the US, yet they keep charging ludicrous prices for it. Softlayer's prices on bandwidth have actually increased! WTF?
So S3 for any decent site makes little to no sense to me, unless you just have money to burn.
Moscato
Active member
That's where it doesn't make sense to me either.
If you're running a big site, the bandwidth gets so expensive it's completely absurd.
Literally the only thing that can be considered a "hack" here is scripting your auto scaling by hand.
AWS pricing gets stupid, to the point where it's detrimental to your own business, very, very quickly...
As to cloudflare... new york times, and reddit uses it btw
Jim Boy
Well-known member
A small but incredibly impottant point for us. Having autoscaling fail at a crucial time, which is the time it would most likely fail, would cost us alot more than the savings we would make by running on Linode.
Jim Boy
Well-known member
and both those organisations use AWS
Moscato
Active member
With cloudflare, you aren't subject to the possibility of a bandwidth $$ attack
Minecraftly
Active member
You can use CloudFlare with AWS.
In fact, if you use CloudFlare, then it's much better to use with AWS than with Linode. CloudFlare & AWS are directly peered, and live within the same data center in most locations (Mostly Equinix using Equinix Cloud Exchange).
In regards to additional cost savings, CloudFlare recently announced partnership with Google Cloud to further reduce GCP bandwidth cost using CF as front end proxy. I personally feel they will do the same with AWS.
P/S: If you don't want AWS then you don't want AWS, there's nothing wrong with it. But comparing the whole AWS product offerings with Linode is never going to be apple to apple.
Last edited:
Jim Boy
Well-known member
I'll let you into a little secret. everything in the internet is scripted!
You might as well be saying that I shouldn’t''t use an O/S, because that's just a bunch of scripts.
OK, so I riight now I make autoscaling decisions based upon latency at the load balancer as well as CPU usage. Yeah it's doable, But I need to be able metrics diligently and reliably because I need to scale before things stop working by looking at trends, not just one-off measurements, so I need to store those measurements, not that node balancers appear to provide such metrics, so some sort of database is required. Also I need to run two separate servers to run the autoscaling scripts - I can't afford downtime remember so some redudancy is needed. Those scripts need to be rock-solid and reliable. Money for this is going out the window while I do this, plus I my own time doesn't come cheap. I'm pretty confident in my own abilities, but AWS have 250 separate teams working full time on improving AWS, and unlike others, I do have a life and can't be bothered reinventing the wheel.
Moscato
Active member
Requiring 2 database servers is a strawman.
You don't need a database server to track your load, you can do it entirely in memory on the servers themselves using 5 second cpu and network polling. The amount of processor to do that is minimal.
If you feel it entirely necessary, you can have 1-2 super tiny instances centrally managing the data (what little there is), or you can just make it distributed across all of the nodes.
You can use 2 way keep-alives between the node balancers and the application servers. If either of them go down, spawn a new one.
This sounds about 1- days of work. It's definitely a job for a programmer though, and not someone who just thinks running websites is cool.
You don't need a database server to track your load, you can do it entirely in memory on the servers themselves using 5 second cpu and network polling. The amount of processor to do that is minimal.
If you feel it entirely necessary, you can have 1-2 super tiny instances centrally managing the data (what little there is), or you can just make it distributed across all of the nodes.
You can use 2 way keep-alives between the node balancers and the application servers. If either of them go down, spawn a new one.
This sounds about 1- days of work. It's definitely a job for a programmer though, and not someone who just thinks running websites is cool.
Similar threads
- Locked
