Neverfail

Teledude

Teledude

Teledude

Soldato
Joined
4 Dec 2003
Posts
2,847
Anyone used it? Opinions?

I have a consultant due in Monday to check our enviroment and demo it but not sure why our project leads are suggesting it when we have mirrored SANs and ESX clusters :confused:
 
Are your systems mirrored across sites? Do you need app aware replication and failover? If so then mirrored sans and ESX clusters might not give you want is required
 
I've got neverfail for exchange. What information are you looking for exactly (ESX module could be different to what I'm used to)?

It basically clones a system to another hardware box which is masked from the network, I'm not sure if you'd call it a hot or cold standby, but it 'should' fail over automatically in the event of a set of defined parameters, normally a hardware/service failure on the primary/live box.

This help?
 
J1nxy said:
Are your systems mirrored across sites? Do you need app aware replication and failover? If so then mirrored sans and ESX clusters might not give you want is required
Click to expand...

There isn't much you can't do with a decent SAN these days, I'd much prefer to stick to the SAN replication than stick in another software package to complicate things more...
 
There is actually still quite a lot you can't do with SAN replication especially when it comes to bringing the hot failover system online automatically.
 
I'm not saying it is a given but there are many situations where SAN based replication does not meet the requirements for the application or data protection of the customer.

I have customers which use SAN replication, products like Neverfail and native replication in the same environment as thats what meets the RTO and SLA requirements. Pure SAN based replication would not meet their requirements.
 
Its good, but i found it very vulnerable to network conditions.

If the customer screws the network up (like ours did) neverfail can (did) go into some horrible split brain situation where both sides try to be the live one. Caused all sorts of problems that one.
 
Stolly said:
Its good, but i found it very vulnerable to network conditions.

If the customer screws the network up (like ours did) neverfail can (did) go into some horrible split brain situation where both sides try to be the live one. Caused all sorts of problems that one.
Click to expand...

But to be honest if they manage to screw the network to that extent it won't matter what product you have it'll more than likely end up in that situation.

Microsoft for instance recommend that the FSM for a Geo-CCR reside on a 3rd site if possible to stop this happening.
 
I'd wait for VMware to release their Fault Tolerance option for ESX rather than a 3rd party option to do the same or similar function. Of course, the VMware option could be hideously expensive :)

The demo is very impressive, it's essentially two VM's running in processor lock-step. If one fails it's offlined and the other one takes over. No modifications required to the VM's at all.

There's a video of it in action here:

http://download3.vmware.com/vdcos/demos/FT_Demo_800x600.html

-Peter
 
Stolly said:
Its good, but i found it very vulnerable to network conditions.

If the customer screws the network up (like ours did) neverfail can (did) go into some horrible split brain situation where both sides try to be the live one. Caused all sorts of problems that one.
Click to expand...

You may want to let them know there's a configuration page called 'Split Brain Avoidance', (going from memory) and to have this confugred.
 
I wouldn't touch never fail with someone elses barge pole. I've seen the fall out when it got an exchange fail over very very wrong and sufice to say it wasn't pretty at all. the best solution I've used was mirrored Netapp boxes at different sites with a live cluster and a single node standby cluster. the time to switch bettween the two was less than 15 minutes.
 
Funny you should mention this, I recently had a very bad Sunday evening after performing a test failover the Saturday before.

All the information stores did not sync correctly, leaving me with 7 corrupt and unmountable databases. Getting things back to normal Sunday PM (and well into Monday am) was a long job, the things you do for the business :)
 
mr.bond said:
Funny you should mention this, I recently had a very bad Sunday evening after performing a test failover the Saturday before.

All the information stores did not sync correctly, leaving me with 7 corrupt and unmountable databases. Getting things back to normal Sunday PM (and well into Monday am) was a long job, the things you do for the business :)
Click to expand...

Sounds familiar, it was a scare I could have done without so I'd never touch one of there products again, I'm also very weary of anything that automatically fails over to another copy of a database stored elswhere. I don't mind having one button to click to automate the failover I just like being in control of when and where it happens.
 
we use it here for Exchange / SQL and File

we all nickname it Always Fail or Neverwork !!

It's a pita really and have never switch over in a live enviroment. We plan to move all our critcal servers over to ESXv3.5 shortly.
 
You must log in or register to reply here.
Back
Top Bottom