December 22, 2008 at 11:28 pm
Me too needs a solution for this. I am also facing the problem.
Thanks
December 23, 2008 at 2:31 am
If i choose Clustering, which should preferable Active /Active or Active/Passive?
December 23, 2008 at 4:56 am
I think in active/passive u dont need to pay for license on the passive server. 😉 but thats only the cost part of it..
December 23, 2008 at 5:08 am
ps (12/23/2008)
I think in active/passive u dont need to pay for license on the passive server. 😉 but thats only the cost part of it..
Thanks,
Can u differentate Active/Passive and Active/Active.
Bac i don't have deep knowledge and confuse to it.
I simply know that in Active/Passive alustering one is active at a time. is it right?
Please expalin..
December 23, 2008 at 5:19 am
I've never implemented it. Wont be able to say much than what's written in BOL. :hehe:
December 23, 2008 at 5:24 am
Paresh Prajapati (12/23/2008)
Can u differentate Active/Passive and Active/Active.Bac i don't have deep knowledge and confuse to it.
I simply know that in Active/Passive alustering one is active at a time. is it right?
Please expalin..
Paresh,
Active\Passive clustering:
you have a 2 node windows cluster with SQL server2005 installed. There are 2 instances which are active on one node, which is node a and fail over to node b
Active\Active:
you have a 2 node windows cluster with SQL server 2005 installed. There are 2 instances 1 active on node a and 1 active on node b. Upon fail over both instances would be active on one node.
This is a basic overview and the more nodes\instances you configure the more complex it gets, should give you the general idea though
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
December 23, 2008 at 6:26 am
Paresh,
I've read this whole thread and agree with a past poster that there is some good but also some REALLY BAD advice here. I really would urge you to seek some professional and experienced advice before setting anything up. Once again it’s nothing personal but to go from a question, LS vs HA/HP Mirror to A/P or A/A Clustering shows a complete lack of understanding for the underlying technologies.
You never answered the first question which was 'What is this for / purpose?', if you cannot answer this (if only to yourself) you have no hope of long-term success and every chance of data loss, downtime and job-seeking.
We know that you have a 10GB db (which is actually quite small) and 500 concurrent users and your HK server does not have users of the app DB (because neither Mirror nor LS use online databases), so what are you trying to achieve? If it’s just a DR copy, then yes LS or Mirror would work. I would go for HP Mirror as HA will require a third Witness server (noticed no-body has even mentioned). LS is not bad for this situation but give you less control and isn’t always reliable (plus very slow and cumbersome).
Clustering is a horrifically bad idea and I'm surprised nobody jumped on it. If you only have Internet links and not point-to-point, this could be difficult to achieve anyway. You also need to have shared storage (preferably SAN) which needs to be accessible (with low latency) from either US or HK. Realistically this won’t be possible with Internet links and might not even work with transatlantic point-to-point. Clustering also has a single point of failure, the storage. Assuming your HK site is your SCF/DR and in a failure of your primary server in US, you want to fail over, your HK server would be accessing your US storage, to then pass data back to the US (unless you have mirrored SAN's, which I doubt), now who else thinks that’s a dumb idea?
Simply put, if it has to be done now, use HP Mirror, BUT and that’s a big but; get some advice, look at your link speeds and point-to-point response times and do a risk/need analysis.
I run multiple A/P clusters (onsite) and mirror most databases to an SCF/DR site half way across the country using point-to-point fibre links, we also have secondary SAN-to-SAN brick level replication (HP CA) and it work great, but even we have issues when bandwidth usage of switch load on those links forces the latency to go up above a certain amount, so who knows what a transatlantic Internet links is going to do.
HTH..
Happy Christmas & New Year.
Adam Zacks-------------------------------------------Be Nice, Or Leave
Viewing 7 posts - 31 through 36 (of 36 total)
You must be logged in to reply to this topic. Login to reply