• Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 3688
  • Last Modified:

How to handle DNS to create failover hosting solution ?

Hi,

This week, 1&1 Internet failed us badly. We lost lots of money, precious traffic while a huge marketing campaign was running in our city. Our web site went down for 3 "strategic" days of our campaign and I've learned (painfully) that 1and1 internet does not have ANY fail over redundancy hosting servers.

This is why I am here.

As a positive nature, I look forward and aim to implement a solution that will prevent such failure in the future.

Here is my plan and I wish to get councils from GURUs like you guys.

1. My intention is to subscribe for another hosting plan but with another web hosting company (maybe GoDaddy) as we need MS Hosting exclusively.

2. After I plan to handle the switch at the domain level.

The question is, how to make sure that when one hosting provider screw up, the other will take the relay flawlessly ?

a. I thought at the first place that it would be possible at the NS settings level. I would set my domain 2 first NS settings to the main hosting company and then the 2 others to the second hosting company.
Question: Is this possible ?

b. My other option is to purchase a No-Ip plus service and handle the domain with redirection only. No-Ip provides a fail over solution. If one goes down, the other IP will take the relay, no matter where I reroute the traffic.
Question: Would it be easier this way ?

I guess knowing 1&1, GoDaddy and No-IP services would help to provide an answer about this one. ;-)

Others may suggest other avenues that I may consider and evaluate.

Thanks in advance !
0
SirTKC
Asked:
SirTKC
  • 4
  • 4
3 Solutions
 
Chris DentPowerShell DeveloperCommented:

> The question is, how to make sure that when one hosting provider screw up, the other will take
> the relay flawlessly ?

In DNS? You can't unfortunately. At least not automatically or without changes in DNS.

DNS has no application level awareness except for DNS itself, meaning it will happily failover on NS records, but every other record type is down to the client. For Host or A records that's simply down to Round Robin, rotation of the order of the responses, the client itself always uses the first in the reply.

Failover in DNS typically works by modifying a Host record for a service. That is, if a service is discovered to be down, the record will be modified to point to a new host. As this relies on record modification it is important that the TTL (Time to Live) for the record is also short. Remote DNS servers will only look for a new entry after the TTL for the current entry (in cache) has expired.

This is an example of a company that runs a DNS failover solution:

http://www.autofailover.com/

HTH

Chris
0
 
SirTKCAuthor Commented:
Hi Chris,

Probably I've asked the question the wrong way.

Can I enter for example in my Domain NS server settings

NS1.HOSTING1.COMNS2.HOSTING1.COMNS1.HOSTING2.NETNS2.HOSTING2.NET
Of course in my example I have 2 hosting company (HOSTING1 and HOSTING2)

So this way, the NS settings are permanently set at the registrar level (independent of the hosting services - of course). It should simply point to the shortest TTL, providing by the same way, the best users experience ?
0
 
Chris DentPowerShell DeveloperCommented:

Yep, you can set it like that. But it won't help in the event of web server failure.

Round Robin is used to return NS records to a requesting name server, the requester will simply use whichever is first in the list, only failing over if the Name Server it's attempting to ask is down. You cannot control which name server a client / requester will ask.

That leaves you in a bit of a poor position if the server hosting the web site via hosting1.com fails. Both name servers for hosting1 will continue to give out the bad IP address (for the failed server). You'd need both name servers to completely fail to get any kind of complete fail-over to hosting2.

From what you've said, I would recommend that you either invest in a host that can provide server / site level fault tolerance (which is bound to cost), or invest in one of the solutions that modifies the DNS record should a site fail (which also carries some cost in addition to the second host).

Your position is not an uncommon one, or at least it's not if I'm understanding it correctly. Unfortunately, DNS is a poor place to attempt to implement fail-over, your choices are very limited.

Chris
0
Concerto Cloud for Software Providers & ISVs

Can Concerto Cloud Services help you focus on evolving your application offerings, while delivering the best cloud experience to your customers? From DevOps to revenue models and customer support, the answer is yes!

Learn how Concerto can help you.

 
SirTKCAuthor Commented:
Hi Chris,

I have attached what I found on our No-Ip Plus services. I believe it refers to what you described above.

The thing is, it ask top enter an IP Address. I have attempted to connect to 1&1 Internet by using IP address only, and it returns "Bad Request (Invalid Hostname). So it has to come under the form of a domain name.

I am quiet confused. :-/

no-ip.gif
0
 
Chris DentPowerShell DeveloperCommented:

That'll be the web server using Host Headers to filter requests to a specific site when hosting multiple sites on a single IP (you can host hundreds of sites on a single server using that).

Round Robin though is another of those that couldn't care less about the state of the service it's providing an IP for. It provides a simple form of load balancing, but as it's not aware of the service state it will continue to hand back "down" IP addresses should one of the servers fail.

In the above, you can enter an IP, as long as the name used to access the IP matches the name used to access the site.

Hope that makes sense...

Chris
0
 
SirTKCAuthor Commented:
Ho it makes a lot of sense what you explained all the way through.
At least it lightened a lot of bulbs !

My next step will be to have a good talk with these guys from No-Ip and see what could be done.

By the way, just to let you know, 1and1 Internet or 1&1 Internet, as you like, still have not restored our web hosting. Apparently they're still rebuilding the raid. I moved all our stuff on Thursday and had it running under another package with them in 3 hours...

The joke is, this morning they sent me the invoice and expect me to pay the full amount for the failed service... huhuhu... Crooked like that I love it...
0
 
SirTKCAuthor Commented:
Thanks for the time spent on this one Chris !!
0
 
Chris DentPowerShell DeveloperCommented:

Ack, that is a bad situation. Been on 1&1's side of that as well, although I like to think my response to restore service was rather faster ;)

Good luck getting this all sorted out :)

Chris
0

Featured Post

Free Tool: Port Scanner

Check which ports are open to the outside world. Helps make sure that your firewall rules are working as intended.

One of a set of tools we are providing to everyone as a way of saying thank you for being a part of the community.

  • 4
  • 4
Tackle projects and never again get stuck behind a technical roadblock.
Join Now