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.
Comments
Nothing is 100%, but stuff like AWS and even iwstack is close to 100%.
Nothing will ever be 100%. We have a high-availability cloud VPS lineup starting at $6/mo - with fully redundant compute, storage and network which is appropriate for applications where uptime is a priority.
https://cc.delimiter.com/cart/cloud-resource-pool/
no, when hosts say 100% SLA they just mean that they'll provide credits/refund if they fail to meet it
iwStack should be good for this though like linuxthefish said
Surely it can be 100% between periods when nothing goes wrong.
Make sure you have your own redundancy as well as provider redundancy, run a ton of DNS/web/database servers, and make sure your setup can tolerate failures without errors.
No honest server provider, physical or virtual, is going to guarantee you 100%. Even with massively redundant everything with geographical independence for some of the redundant parts there will be possible faults that affect availability.
You can get progressively closer to 100% with good redundancy design in your infrastructure and your application but it takes knowledge and care design & planning. Good cloud providers are experts in the infrastructure parts, and the software parts too if their offerings fit your application precisely enough, but they have all had problems in the past that took out chunks of their infrastructure temporarily. Also: you still need to properly design your application to take full advantage of what they offer. Massive concurrent power or network faults take out a DC temporarily: did you pay for multi-DC redundancy? Fault tolerance might keep the platform running as everything switches between DCs, but does any of your application logic fall over when the switch happens?
That is a very generic answer, if tell us what you are doing that requires high availability someone might be able to give you more specific pointers as to how you might get as close to 100% as practically possible with current technology and services. Also a rough budget would make a significance to the advice you might get. Trying to guarantee 99.999 (about 5 minutes down in one year) costs a heck of a lot more 99.99 (52 min/yr or 4m20s/month) which in turn costs a lot more than 99.9 (about three quarters of an hour per month). People often change their mind about absolutely needing five nines when they see the quoted costs...
No one is going to say really 100% uptime. Just because you never know what will happen. But more often than not you will find some providers that have better uptime. I would recommend looking at hostdime.com
AWS goes down all the time
Just courious, how can google have 100% uptime? Whats the tech behind it?
Google has not 100% uptime
In order to really negotiate a true 100% SLA both sides first have to agree that it is impossible to guarantee it, but the Service level Agreement is based on achieving 100% within pre defined parameters subject to penalty, for example nuclear strike would not count towards down time.
The reality is though while you could fine a company such as softlayer who will negotiate on this level the price would allow you to design your own farm of 10 servers in 10 locations with geo dns or anycast that would actually out perform softlayer in terms of redundancy.
Keep in mind that even IWStack had outages lasting a day or so in the past.
@Op: The best thing is probably to not just rely on one server but to to build redundancy using AWS or something similar.
How can hostingspecialist offer 100% uptime lol?
http://s15.postimg.org/fyos2et1n/image.png
I belong to a few forums that fell for the host on AWSAnd never go down sales pitch typical sales droid speak is what it proved to be.
Google is not just a website. You approach a different server if you are in London, Seattle or Tokyo. Even from the same town, you recall often a different server from google. It is a different infrastructure and even with this multimillion infrastructure, Google has not 100% uptime, it's just won't face downtime to all of their servers and if the first server you are going to reach is down, their infrastructure will lead you to the next available. And, yes, Google did have some downtime to whole regions (massive downtime last February).
But don't compare google to a rented server. It's like compare transportation with a bicycle compared with NASA's spaceships.
Yeah bicycle is a lot more reliable.
I will just leave this here (RamNode)
Having 367 days uptime, it doesn't mean that you will get 10 or 20 years without single downtime, as vps/dedicated server at some point will need to update the kernel or your provider get issue with network.
Nice that your server was up for 300+ days. What about the network? (No real question, don't answer)
Your server can show 3,000 days of uptime and still the network can go down every 5mins. So this metric is just a part of the general uptime.
They're offering a guarantee, meaning they will probably give you a few cents every month if you don't forget to claim it and are eligible for it if downtime happens.
You were bombed with a sales message.
No need to claim it, it will be added regardless. As @GM2015 stated, @WHT, we were offering you a guarantee. As I also said, we can offer near enough 100% uptime based on the redundancies I spoke about in the PM and our DDoS protection. I'm speaking based on our previous 6 months at this datacentre. All 3 of our servers haven't been offline since they were racked.
I wouldn't call it bombed either. I sent one message relating to his question, offering my more affordable solution.
They are probably industry leaders or something like that which enables them to promise the impossible.
Good joke, but they're selling a guarantee. It's like selling a dream.
Yes, we can twist and torture the language the way we see fit.
Is that you ignite again?
It's more likely to be our friend Nuggets.
Correct. I've seen Google going down every now and then either due to bad deploys or someone fucking up the network configuration (BGP routes suddenly disappearing is one of the most things I've seen so far).
industrial standard uptime is golden nines.. 99,999%. there is no 100% if you still depend on machines
That's just stupid.
As others have pointed out, it's not just the server, you also need to have equal performance from the network.
99.99% is down less than 52 minutes per year.
99.999% is down less than 7.4 minutes per year.
Its common sense... Do you know future.. No.... Then how would the provider know if for example hdd fails... Tommorow.... 100%...DNE... In real world.... Just get two completely differnent setups and switch if one fails.... And just hope that both dont fail at the same time :-)