How to have 100% uptime

[dcb1101]

I know most will say its not possible, but how do you get as close to 100% as possible? Assume good server with good host, how do you make sure you will always be online?

 

[RDO Servers]

You are correct in that there is no way to guarantee 100% uptime. There are always points of failure and even the possibility for cascading failures.

With that said there are thing you can do to decrease your uptime.

1. Like you said, make sure you have a quality server, with a quality host, at a quality data center with its own redundant systems.

2. Add website monitoring via a service like pingdom so that you can catch outages fast. Having your site go down is bad. Not catching it for 8 hours, is even worse.

3. Server occasionnaly hang up or need to be restarted. Adding a second server behind a load balancer will take are of these issues. If one goes down, the other server will still serve traffic. This is also use full in case of load spikes.

4. Load balancers can also have problems and using one, creates a single point of failure. Ideally, you will want 2+ load balancers for 2+ separate clusters. Now if a LB or entire cluster goes down, the other LB cluster will still server your sites.

5. Although it is rare at quality data centers, data centers do experience problems. Fiber cut, natural disaster, main router, etc. A good DC should be fully redundant, but you can have degraded service while they are running on backups. (600GBs fiber gets cust, so there running on the backup 200GBs line, like happened to OVH recently). To solve this, you need to have fail over across multiple data centers.

6. Using DNS fail over is the easiest method, but you can still have some downtime if things don't switch over as they should. There are 3rd party services that offer DNS management with immediate data center faukiver, but it get pricy!

Long story short, there are a lot of things you can do to dramatically decrease your downtime, but you will need some deep pockets!

We have a couple customers with load balanced clusters, but we always makes sure they know that the is still the possibility for downtime as noting is ever foolproof!

 

[novandak]

novandak

How to monitor my web hosting up time via tools like that?

First time ever heard about this

 

[RDO Servers]

RDO Servers

[URLnf="https://www.pingdom.com/free/"]Pingdom free website monitor[/URLnf] will email you if your site goes offline.

 

[Hassan]

Nothing is impossible you need find good host with professional team.

 

[dcb1101]

dcb1101

how do you provide 100% guarantees? Does your hosting have all of the failsafes? Do you have a uptime record somewhere?

 

[Hassan]

Yes we do provide 100% guarantee you can google reviews about us its just that everyone is in hosting business now a days very rare are professionals i offer you try our services for free and find how we can guarantee.

 

[RDO Servers]

RDO Servers

I believe he was asking at the possibility of having actual 100% uptime, not just SLA that guarantees it and give account credit if there is a outage. I don't doubt your services at all, I'm sure it's great, but no one can truly guarantee 100% uptime.

Even major corporations and social networks have experienced downtime, and they have more infrastructure than most hosting companies. It's just a fact of life.

 

[Hassan]

Hassan

Yes its possible to provide 100% uptime with Tier IV datacenter due to lack of knowledge most of us around here not familiar with technologies i can understand major corporations such as EIG group have downtime there are several reasons behind it they don't want to spend much money on network infrastructure and hardware as well they want to maximize profit many other reasons behind it i cant discuss everything here as its not related to post but anyone who want to learn can contact me.

 

[Hostlumina]

Hostlumina

I do not think it's lack of of knowledge from anyone. The reality is this.... nothing is 100%.

As mentioned before there is always a fine print that the 100% SLA will not cover (e.i hardware). You could have the best of the best redundant network but something could happen. Keyword "could".

I would love to see a TIER IV datacenter advertising 100% up-time... They still leave room for about 0.8 hours of interruption.

 

[Hassan]

Hassan

0.8 hours of interruption/year Availability: 99.995%

 

[Hostlumina]

Hostlumina

Correct. They do not provide 100% up time of their network. They know something "could" go wrong giving them 28 hours a year.

I would love to see 100% up time provided with zero restrictions in the contract. I will sign up right now. D:


It be like buying a new benz and the dealer offering "100% mechanical free" for the life of the car. It will never die on you nor need repairs. Just put gas on it and go go go. :scooter:

 

[RDO Servers]

RDO Servers

In 2014, Amazon Elastic Compute Cloud experienced 20 outages and had a total uptime of 99.9974 for the year. ([URLnf="http://www.networkworld.com/article/2866950/cloud-computing/which-cloud-providers-had-the-best-uptime-last-year.html"]http://www.networkworld.com/article/2866950/cloud-computing/which-cloud-providers-had-the-best-uptime-last-year.html[/URLnf])

I think we all would love to know how your servers are more reliable then Amazon's.

Is this too a result of Amazon's "lack of knowledge most of us around here not familiar with technologies"?

 

[ElixantTechnology]

ElixantTechnology

Even the biggest and baddest networks will see hiccups. For example, even Facebook/Instagram see outages from time to time, it's how they are handled that is what truly matters. This is where a 100% uptime SLA comes into place. More or less a sales pitch, but in the end, customers aren't looking for hosts that claim it, that is why we have lowered our SLA to 99.995%, because in the end something *could* in theory happen.

Most of the Amazon outages were planned, being because of updates that require a reboot to the core of Xen Hypervisor, and maybe some unplanned ones as well. Ultimately the best thing to look it as how your host handles the circumstances of said outages. Make sure you are with a provider that keeps you informed about any issues that arise for one, actually, that is the biggest thing to keep look out for.

 

[Hassan]

There are many tools such as phpservermonitor.

 

[Hostlumina]

Well, it is possible for a server not to lose any power or network. However the longer it runs the higher chance of something happening down the road.

If you find a provider offering such. I would suggest reading through the legal notes and fully understanding them. You will find that usually the SLA will not apply for certain reason (this usually applies to every host include us).... so is it really 100% up time? or a marketing thing?

 

[ElixantTechnology]

You can offer customers a 100% uptime guarantee on the grounds that 99.999% of the year your servers is online 100% of the time, can be confusing at times, however it is possible. Utilizing a Datacenter that operates redundant power feeds, redundant cooling and redundant network uplinks is the first step. Next, utilize a server that operates with redundant power supplies, a RAID configuration that allows for how-swap hard-drives in the event of a failure and software that allows for the live-update of kernel modules and important software without the need for reboots.

By doing so, with proper configurations you can allow for near 100% uptime as stated intially in my message. For example, some hosting providers allow for 99.9% uptime guarantee, which is open for up to 7 hours of downtime per month. Over the last year, our servers have seen a total of 2 hours of downtime (before the redundant upgrades took place). A SLA covers Hardware, Network, Power and Cooling. Software on the other hand is not covered in the fine-print, therefore the occasional restart of Apache, MySQL or other services (which should only last 1-9 seconds) is permitted in the event of software tweeking taking place to increase customer satisfaction in quality of service.

 

[Hassan]

I am ready for contract pm me for discuss as its not related to thread thank you.

 

[Hostlumina]

Hostlumina

We can keep the discussion open here, I think it is related to the 100% up time guarantee.

I am sure many would love to learn how 100% can be guaranteed including myself. How can we keep a web site 100% up and running even in the event of hardware malfunction, all hard drives die at the same time, datacenter cables cut (yes physically cut during construction near by it has happen), connection issues due to earth quake or floods across the country destroying 30% of the datacenter, guarantee that a 400 gigabit per second DDOS attack to a site will remain 100% up and running and not affect anyone else within the network, and so much more...

Yes some of the above might be excessive scenarios but let's face it...it "could" happen. Even in a cloud environment.

In a DDOS scenario of 400Gbps. I bet the SLA agreement + TOS would then be brought up stating that due to such a big DDOS attack, the site was taken offline and no compensation can be given. Can the sites be accessible? Yes, but only by the owner most likely until the attack is done with. To the customer the site was probably not 100% up.

edit: FYI typing from mobile not so easy. sorry for any typos in advance.

 

[dcb1101]

Now this is a discussion! Great to see all the honest feedback of whats actually possible instead of just sales pitches!

 

[Hostlumina]

Hostlumina

For sure. As long as is kept civil it makes a great discussion. :icon_mrgreen:

 

[ElixantTechnology]

When selecting a Hosting Provider understand that even if they over a 100% uptime guarantee, things *can* happen as others in this thread have said. The real question you should be asking each and every provider, is what steps they take to prevent such outages and the steps they take during/after something occurs.

 

[Hassan]

Hostlumina i dont think so you even visit Tier IV datacenter i believe on results rather than argument you offer me lets contract i will provide you 100% uptime how thats my responsibility.

 

[Hassan]

You need find info about Tier IV datacenter , Raid technology, IP Failover and even if cables cuts there is backup via other lines and wireless.

 

[MightWeb]

MightWeb

You need to understand that it doesn't matter how many redundancies you have. Even if you build failover systems, it takes time for the node to realize that a server is not responding, and when it's not, it takes time for the failover to engage. Remember in 2013, when Google had 5 minutes of downtime? Is that due to their lack of a quality datacenter, failovers and multiple bandwidth providers?

You can come really, really close - but you can not get 100% over a longer course. It is not possible, no matter if you're in a Tier Star Trek Enterprise-datacenter.

 

[Hostlumina]

Hostlumina

Is there any Tier Star Wars Enterprise-Super-Dooper-mega 100x redundant-datacenter? I need to make sure I visit it ASAP.

All jokes aside. Right on the spot. The longer you run the more likely something "could" happen. You can be close as you said it.

 

[Hassan]

Hassan

The IP Failover may be moved from one server to another in a few seconds. i can understand what you are saying but we are talking about normal conditions.

 

[RDO Servers]

RDO Servers

Yes, IP failover takes a few seconds and yes this is normal. The point we are making is if your site is down for a few seconds, that is NOT 100% uptime!

 

[Hostlumina]

Hostlumina i dont think so you even visit Tier IV datacenter i believe on results rather than argument you offer me lets contract i will provide you 100% uptime how thats my responsibility.

?

You just don't understand that nothing is 100%. Something "could" happens as everyone else has stated. You can come close but something "could" happen in the long run specially the longer you run.

I highly doubt you will guarantee anyone 100%. Your web site also only offers 99.9% and also includes conditions (which is normal nothing wrong with that) protecting you (restrictions - not no restrictions).

Your TOS:

Credit shall not be provided to YOU in the event that you have any outage resulting from:

- scheduled maintenance as posted from time to time at PostPayHost Inc,
- your behavior or the performance or failure of your equipment, facilities or applications,
- circumstances beyond PostPayHost Inc's reasonable control, including, without limitation, acts of any governmental body, war, insurrection, sabotage, embargo, fire, flood, strike or other labor disturbance, interruption of or delay in transportation, unavailability of interruption or delay in telecommunications or third party services, including DNS propagation, domain name registration/transfer, failure of third party software or hardware or inability to obtain raw materials, supplies, or power used in or equipment needed for provision of your web site,
- YOU breaking any agreement policy in PostPayHost Inc's "Terms & Conditions and AUP" causing a machine to fail as a result.

 

[Hassan]

Hassan

TOS have many legal terms you cant say as per TOS no one can provide 100% uptime i have good reviews you can google them.

 

[Hostlumina]

Hostlumina

But the TOS is protecting you from providing 100% up time in "case" something happens . If you can provide 100% guarantee up time as you have been stating why not remove it and advertise 100% up time?

You said earlier and I quoted you stating to PM you to do a contract when I mentioned that I would sign up if some provided 100% up time withzero restrictions (not necessary truly going to sign up why? cause i don't trust it. my personal opinion)

 

[Luca]

100% uptime is possible when we have acquired the level of knowledge required to create a space that is outside the regular time-space continuum. This space would be void of anything except things that are needed to run the datacenter, that way, no external factors "could" happen to the datacenter physically.

We also need to device a way to perpetually feed this space with energy required to run the datacenter, and a way for the datacenter to available for you no matter where you are. And to do so, we need this space to be omnipresent across the universe, and intelligent, so it can gather energy by opening wormholes near the billions of stars out there to feed its energy needs. Being omnipresent also means this hypothetical datacenter can detect where data access are being requested, and open up mini wormholes to accomodate them.

This datacenter would be maintained by AI's that are infinitely smarter than human beings, which of course will upgrade themselves as time goes on.

In no time, they will be godlike beings who are omniscient, omnipresent, and omnipotent - that still provide us our puny little 100% uptime datacenter because they don't see the need to shut it down, considering how of little significance it is to them.

 

[RDO Servers]

RDO Servers

Maybe thats just what Hassan has done.
Seems fitting considering his avitar.....

Hassan, are you a Cylon?

 

[Hostlumina]

100% uptime is possible when we have acquired the level of knowledge required to create a space that is outside the regular time-space continuum. This space would be void of anything except things that are needed to run the datacenter, that way, no external factors "could" happen to the datacenter physically.

We also need to device a way to perpetually feed this space with energy required to run the datacenter, and a way for the datacenter to available for you no matter where you are. And to do so, we need this space to be omnipresent across the universe, and intelligent, so it can gather energy by opening wormholes near the billions of stars out there to feed its energy needs. Being omnipresent also means this hypothetical datacenter can detect where data access are being requested, and open up mini wormholes to accomodate them.

This datacenter would be maintained by AI's that are infinitely smarter than human beings, which of course will upgrade themselves as time goes on.

In no time, they will be godlike beings who are omniscient, omnipresent, and omnipotent - that still provide us our puny little 100% uptime datacenter because they don't see the need to shut it down, considering how of little significance it is to them.

:afro1:

Maybe thats just what Hassan has done.
Seems fitting considering his avitar.....

Hassan, are you a Cylon?

I am going with Sonny from i.robot.

 

[elcidofaguy]

 

[Hostlumina]

Hostlumina

Enough said right here! All questions answered.

 

[dcb1101]

dcb1101

Bwahahahahahaha

:rockon:

 

[Hassan]

Hassan

Laughing out loud

 

[sharplesson]

If you are a web hosting provider, then yes you can handle uptime for your clients. But if you are just a blogger, then how come you are dealing with up time? Because it is nothing to do with you.

But yes, you can do some tricky things to save bandwidth, web space and improve your page load time.

thanks
web hosting tips
sharplesson

 

[aschatria]

A reliable hosting provider will give you high uptime but I am not sure about 100%. I am hosting my pages for free on Blogger and never had any uptime problems. Doe with other hosting ( even paid) I heard a lots of complaints.

 

[lorettalschmitz]

If you can't afford cloud, you definitely can't afford replication over the pond (maybe if you have small updates/inserts). Requires layer2/3 switching.

I suggest you house your DB and web servers [Edited by Moderator] where you have the most traffic and serve from there and all static assets (images, js, css, etc) via affordable CDN with the most points of presence in your biggest markets. Have your code determine which imgsrc CDN to render in html depending on the TLD or geoip.

It's definitely easier to rsync static assets rather than replicate databases.