Training Kit Question
-
Contoso Traders Inc. is a financial services company located in Denton, Texas. SupraTrade is the new online trading platform being deployed for the company's 65 million worldwide customers. This platform lets customers do research and place buy/sell orders for any financial instrument (for example, stocks, bonds, options, and futures) on any exchange in the world.
Because the system supports trading on all exchanges around the world, the system must always be available, outages need to be made transparent to users, and performance is critical.
Which availability technology can the company implement to meet all Contoso's requirements?
A. Replication
B. Database mirroring
C. Failover clustering
D. Log Shipping
I gave my answer as C. Failover clustering. What key piece of information makes B the best answer?
When I re-examine clustering, however, it appears to only solve the problem of reducing downtime due to hardware failures. Does failover clustering typically get paired up with another high availability solution so that downtime due to corruption on the database can be reduced at the same time? Here, there was only one answer, B, ....but do the two get implemented together in real life situations?
-
One of the features of database mirroring in 2008 is automatic page repair which would allow you to restore corrupt pages directly from the copy. If the database were on another server you would also be able to run things like dbcc checkdb on it thereby reducing the amount of time required for maintenance.
Hope this helps,
Rich[p]
[/p] -
Failover clustering is a Windows based solution that offers redundancy for the processor, memory, power supply, network cards, etc, but does NOT offer redundancy on your data. It requires a shared drive for the data to exist on as well as the install directory.
Database Mirroring offers all of the advantages of Failover Clustering (redundancy due to a second server) with the added benefits of data redundancy (the database exists on BOTH servers, offering complete backup in case of hardware failure). Additionally, there is no distance requirement for the Mirroring, as they are separate full systems connected through TCP/IP.
Having Mirroring on a failover cluster (or 2 as the case may be) would be an extremely expensive solution.
Hope this helps.
-
Rich-403221 (7/15/2010)
Failover clustering is a Windows based solution that offers redundancy for the processor, memory, power supply, network cards, etc, but does NOT offer redundancy on your data. It requires a shared drive for the data to exist on as well as the install directory.Database Mirroring offers all of the advantages of Failover Clustering (redundancy due to a second server) with the added benefits of data redundancy (the database exists on BOTH servers, offering complete backup in case of hardware failure). Additionally, there is no distance requirement for the Mirroring, as they are separate full systems connected through TCP/IP.
Having Mirroring on a failover cluster (or 2 as the case may be) would be an extremely expensive solution.
Hope this helps.
Yes, this helps. I believe you've delineated the applicable differences and that it comes down to data redundancy which clustering doesn't offer.
I now understand there is a distance limitation with failover clustering which would make this prohibitive from this environment.
One more question, please! When is failover clustering clearly the best option, then? Since it never directly addresses data redundancy? Is it only when it is used in conjunction with another high availability technology.
-
rdouglas66 (7/15/2010)
One of the features of database mirroring in 2008 is automatic page repair which would allow you to restore corrupt pages directly from the copy. If the database were on another server you would also be able to run things like dbcc checkdb on it thereby reducing the amount of time required for maintenance.Are you saying it is possible to run dbcc checkdb on a mirroring server? Since it is in a recovering state I would have thought running any command against it not possible.
-
this bit makes B the best answer - 'outages need to be made transparent to users'
with failover clustering, when the cluster does failover to another node, there is an outage and the application will have to reconnect. the databases will also have to go through recovery as SQL restarts which can increase the outage time.
Configured correctly and using the latest version of SNAC (the client access components) the application will automatically connect to the mirror if it detects the principal is down, making failover seamless.
with mirroring the data cache on the mirror is also in a 'warmed up' state, which means performance does not take a hit whilst data is read into memory.
---------------------------------------------------------------------
-
Also -
you cannot run checkdb on a mirror database
yes a DR solution such as log shipping or mirroring is often paired with clustering which comes more under the heading of HA, often to a standalone server but sometimes to another cluster.
with windows 2008 you can have a multi site cluster, this is a cluster where the nodes are attached to different SANS, and the drives on the two SANS are kept in line using SAN replication software.
---------------------------------------------------------------------
-
Probably bad form to populate forum with 'thank yous' but very much appreciated.
-
hxkresl (7/15/2010)
Probably bad form to populate forum with 'thank yous' but very much appreciated.no way, its appreciated. Its those that never get back to you to let you know how it went that show bad form. How many times you researched a problem, found a likely thread and it peters out with no final post to say what the resolution was!
---------------------------------------------------------------------
-
hxkresl (7/15/2010)
Probably bad form to populate forum with 'thank yous' but very much appreciated.I certainly agree with George, I appreciate the thanks. Let us know how it goes.
Viewing 10 posts - 1 through 9 (of 9 total)
You must be logged in to reply to this topic. Login to reply