I need to setup a 2 node cluster for testing.
I have 2 identical servers running Win2k3, but I have a question about the
shared disk device.
For the shared disk can I use a NAS device, one that connects thru a cat5,
or do I need a device that uses scsi?
Remember this is for a short term test.
Thanks
SQL Clustering won't support NAS. SCSI would work, but why not try Virtual
Server 2005? That works great!
Cheers,
Rod
MVP - Windows Server - Clustering
http://www.nw-america.com - Clustering
http://msmvps.com/clustering - Blog
"dutch" <dutch@.discussions.microsoft.com> wrote in message
news:2CCAEA46-6F5F-44D6-86A1-5BBAD07F8B18@.microsoft.com...
>I need to setup a 2 node cluster for testing.
> I have 2 identical servers running Win2k3, but I have a question about the
> shared disk device.
> For the shared disk can I use a NAS device, one that connects thru a cat5,
> or do I need a device that uses scsi?
> Remember this is for a short term test.
> Thanks
sqlsql
Thursday, March 29, 2012
Cluster synchronize
I have an architecture which has 4 DB servers on the whole; two will
participate in load balancing and other two in fail over clustering.
My question is how will the two DB servers that is used for load
balancing synchronize?
Hi
To synchronize the data between the 2 clusters, you need to use something
like replication. Clustering is not a load-balancing technology.
Regards
Mike Epprecht, Microsoft SQL Server MVP
Zurich, Switzerland
IM: mike@.epprecht.net
MVP Program: http://www.microsoft.com/mvp
Blog: http://www.msmvps.com/epprecht/
<mukesh.thiru@.gmail.com> wrote in message
news:1135764099.960391.227140@.g43g2000cwa.googlegr oups.com...
>I have an architecture which has 4 DB servers on the whole; two will
> participate in load balancing and other two in fail over clustering.
> My question is how will the two DB servers that is used for load
> balancing synchronize?
>
|||Clustering is also not compatible with Windows Load Balancing, so you would
need to use a hardware based load balancer.
Mike
Mentor
Solid Quality Learning
http://www.solidqualitylearning.com
"Mike Epprecht (SQL MVP)" <mike@.epprecht.net> wrote in message
news:eQ%23n4Y6CGHA.620@.TK2MSFTNGP10.phx.gbl...
> Hi
> To synchronize the data between the 2 clusters, you need to use something
> like replication. Clustering is not a load-balancing technology.
> Regards
> --
> Mike Epprecht, Microsoft SQL Server MVP
> Zurich, Switzerland
> IM: mike@.epprecht.net
> MVP Program: http://www.microsoft.com/mvp
> Blog: http://www.msmvps.com/epprecht/
> <mukesh.thiru@.gmail.com> wrote in message
> news:1135764099.960391.227140@.g43g2000cwa.googlegr oups.com...
>
|||mukesh.thiru@.gmail.com wrote:
> I have an architecture which has 4 DB servers on the whole; two will
> participate in load balancing and other two in fail over clustering.
> My question is how will the two DB servers that is used for load
> balancing synchronize?
If you are using SQL server 2005, you might want to consider using read
only shared databases for your reporting . Using a shared file system,
all the nodes can see the same database files which are read only, and
mount the database as read only. You could run reporting quaries
against up to 16 nodes archiving close to 16X speedup.
If your update cycle and reporting cycle do not overlap, switching
between the read write to read only (and back) takes about 20 seconds.
Note that no replication is needed and no storage space is wasted.
Tomer Meshorer
Database Engineer
Polyserve Inc (http://www.polyserve.com)
Email:tmeshorer@.polyserve.com
|||I'd very strongly disagree with that assessment. It isn't a shared file
system, it requires a SAN.
Additionally, just because you point 2 SQL Server instances at the same
database does not mean you get 2X performance improvement, just like
pointing 16 instances at a single database will not give you 16X
performance. You could in fact see your performance degrade. The only way
that your performance would improve is if you have queries running on each
instance that are not competing with each other for disk throughput. If
there is disk contention, your performance will degrade since a single set
of disks are now servicing requests from multiple instances at the same
time.
Read only, shared databases is a nice new feature, but any performance
benefit you may get is very highly dependent upon your query patterns, the
volume of data in your active working set, and the amount of physical data
you can cache in memory on each instance.
Mike
Mentor
Solid Quality Learning
http://www.solidqualitylearning.com
"tmeshorer" <tmeshorer@.polyserve.com> wrote in message
news:1135935720.275938.306680@.g43g2000cwa.googlegr oups.com...
> mukesh.thiru@.gmail.com wrote:
> If you are using SQL server 2005, you might want to consider using read
> only shared databases for your reporting . Using a shared file system,
> all the nodes can see the same database files which are read only, and
> mount the database as read only. You could run reporting quaries
> against up to 16 nodes archiving close to 16X speedup.
> If your update cycle and reporting cycle do not overlap, switching
> between the read write to read only (and back) takes about 20 seconds.
> Note that no replication is needed and no storage space is wasted.
>
> Tomer Meshorer
> Database Engineer
> Polyserve Inc (http://www.polyserve.com)
> Email:tmeshorer@.polyserve.com
>
|||Your scenario is correct on a specific context. Vis you have an I/O
bottleneck
However, this is customer specific.
In cases where you have a CPU bottleneck (usually occurs in DW type of
quaries,which tends to do sequential I/O) the suggested solution will
scale close to linear(at least as observed in internal benchmarks where
I/O was not a bottleneck)
What I am suggesting, and specifically in the context of reporting, is
an alternative way for scaling SQL server.
Currently the only way to scale is up, which become very costly when
you go to the 8,16,32 CPU boxes.
In addition, you cannot scale economically since you move in increments
of 2^n each time you want to decrease your reporting time (and move
exponentially in price)
Note that in both scenarios (scale up to big SMP, or scale out on a
shared file system cluster) if you have an I/O bottleneck you reach the
state as described in your email.
However scaling out is more economically sound (assuming that you do
not have I/O bottle neck) since:
1. It is much cheaper
2. Pay as you go
3. No storage duplication
4. No changes to the database configuration(I.e. no need for
replication).
Best,
Tomer Meshorer
Database Engineer
Polyserve Inc
|||My only point was that you stated that you would get a 16X performance
improvement by using this when you deploy on 16 machines. That statement is
100% false and we both know it. You get ZERO performance improvement. Not
a single, solitary second of improved performance. A query running on the
operational system and taking 10 seconds is going to take 10 seconds to
complete if you are using a scalable shared database model.
Do not mix improved performance with the ability to execute more read only
queries per unit time. It is NOT the same thing and is an incredibly
misleading statement. It is misleading statements like that which lead to
people implementing things and then saying that "SQL Server can't scale" or
"SQL Server can't perform", because they are trying to make a feature do
something that it was not designed to do and never had the possibility of
doing in the first place.
You CAN be increased throughput and you CAN get increased capacity by being
able to run many more of these 10 second queries simultaneously when they
are executed against multiple machines. The ONLY time that you get
increased throughput (more queries per second) or increased capacity (more
queries per second) is if you hae a VERY specific situation where memory or
CPU is the ONLY bottleneck. In those situations, you can get increased
throughput or capacity, BUT, you do NOT and never will get a linear
increase in query capacity. The amount of increased capacity is VERY highly
dependent upon your database design, the volume of data you are working
with, and query patterns.
You cited DW type of queries. There is no such thing. You either have
queries that perform sequential I/O or queries performing random I/O.
Giving it a fancy name doesn't change it, because you run both types of
queries in every operational system. So, let's look at a CPU contrained
query. We have system where we have 4 processors and 4 GB of RAM. Queries
executing against the 400GB of data within the database normally take 15
minutes to execute and pull as much as 3 GB of data. We have isolated the
issue to CPU bottlenecks and moved the system to an 8 processor machine,
thereby doubling the CPU capacity. We observed that the execution time of
this query drops to 11 minutes whereby, we found that memory became the
bottleneck in the system. We doubled the RAM to 8GB and found that the
query execution only dropped to 10 minutes. Both memory and CPU became
bottlenecks at that point, because we were all of a sudden executing more
queries per unit time and 3 simultaneously executing queries would saturate
our memory. This would cause pages to be ejected from the cache (incurring
additional processor overhead). More pages would have to be read from disk
(more processor overhead). We quickly found out that it if we increased the
processor or memory capacity, we would then shift the bottleneck to the I/O
subsystem.
Sure, you can craft testing matrices and benchmarks that will say that
scalable, shared databases will get you a nearly linear scaling. (VERY
important to note that the word is SCALING, NOT performance improvement.)
For every scenario that you come up which exhibits this behavior, I can
change it a hundred different ways to cause the performance to degrade
either by changing the query pattern, introducing a new query, changing the
indexing, increasing the data volume, changing query parameters, etc.
It is a nice technology that has a VERY specific use in an environment
experiencing VERY specific scalability issues. Unless you have done a full
analysis on the system to determine whether it is appropriate, implementing
this technology would be a very big mistake. It most definitely isn't going
to produce a linear scaling and there is a reason that Microsoft is not
giving any specific numbers. (The instant they publish numbers like you
have, there are going to be several thousand people handing evidence that
the findings are not correct.)
This very specific issue is what makes all of the Oracle marketing and sales
surrounding RAC a complete joke. Oracle RAC won't produce a linear scaling
for exactly the same reasons that Scalable Shared Databases won't produce a
linear scaling. If your query pattern, active data volumes, and bottlenecks
meet a VERY specific profile then these types of technologies will get you
increased SCALABILITY. (That does NOT mean you get improved performance.)
Mike
Mentor
Solid Quality Learning
http://www.solidqualitylearning.com
"tmeshorer" <tmeshorer@.polyserve.com> wrote in message
news:1136012125.047011.237960@.g47g2000cwa.googlegr oups.com...
> Your scenario is correct on a specific context. Vis you have an I/O
> bottleneck
> However, this is customer specific.
> In cases where you have a CPU bottleneck (usually occurs in DW type of
> quaries,which tends to do sequential I/O) the suggested solution will
> scale close to linear(at least as observed in internal benchmarks where
> I/O was not a bottleneck)
> What I am suggesting, and specifically in the context of reporting, is
> an alternative way for scaling SQL server.
> Currently the only way to scale is up, which become very costly when
> you go to the 8,16,32 CPU boxes.
> In addition, you cannot scale economically since you move in increments
> of 2^n each time you want to decrease your reporting time (and move
> exponentially in price)
> Note that in both scenarios (scale up to big SMP, or scale out on a
> shared file system cluster) if you have an I/O bottleneck you reach the
> state as described in your email.
> However scaling out is more economically sound (assuming that you do
> not have I/O bottle neck) since:
> 1. It is much cheaper
> 2. Pay as you go
> 3. No storage duplication
> 4. No changes to the database configuration(I.e. no need for
> replication).
> Best,
> Tomer Meshorer
> Database Engineer
> Polyserve Inc
>
|||I should have clarified my point. I am looking at the actual wall clock
time that it takes to run a reporting job. I also assume that you have
more queries than servers, which I think would be the case for a
typical reporting job.
So from a wall clock time perspective, you will be able to get close to
X#server execution time improvement. I.e. if you are CPU constrained, a
6 hour job with 6 queries on one server can be reduced to 1 hour job on
6 servers.
If you have an I/O bottleneck there are two cases. If you are storage
constrained that this is a natural case. Vis, it does not matter if you
have 4*2 way servers or one 8 way server. If you are internal I/O BUS
constrains (e.g. PCI Express) moving to 4*2 way might improve your
situation since now you have 4 internal buses in your disposal.
However, I do not agree with your other point that this is very
specific situation. I think that you will encounter this situation at
the same probability that you will encounter any other kind of
bottleneck (which is about 1/3 of the cases). However, as your example
describe, to move out of the "CPU as a bottleneck" state into "I/O as
bottleneck state" you have to scale up. Based on your example, your
customer had to scale up from 4 way to 8 way.
Now lets say that you paid the money and now you are I/O constrained.
Ok, you buy more disks, change your RAID method, etc. Time passes and
you are again constrained by the CPU... are you going to move to 16
way?, 32 way? 64 way ?
My point is that if you do not have CPU bottleneck now, you will be
after you solve your current bottleneck. I.e. this is not VERY specific
scenario but rather a common one.
Tomer Meshorer
Database Engineer
Polyserve Inc (www.polyserve.com)
|||So that means that I can put 9 women to work and produce a baby in 1 month?
Mike
Mentor
Solid Quality Learning
http://www.solidqualitylearning.com
"tmeshorer" <tmeshorer@.polyserve.com> wrote in message
news:1136319106.372823.46070@.o13g2000cwo.googlegro ups.com...
>I should have clarified my point. I am looking at the actual wall clock
> time that it takes to run a reporting job. I also assume that you have
> more queries than servers, which I think would be the case for a
> typical reporting job.
> So from a wall clock time perspective, you will be able to get close to
> X#server execution time improvement. I.e. if you are CPU constrained, a
> 6 hour job with 6 queries on one server can be reduced to 1 hour job on
> 6 servers.
> If you have an I/O bottleneck there are two cases. If you are storage
> constrained that this is a natural case. Vis, it does not matter if you
> have 4*2 way servers or one 8 way server. If you are internal I/O BUS
> constrains (e.g. PCI Express) moving to 4*2 way might improve your
> situation since now you have 4 internal buses in your disposal.
> However, I do not agree with your other point that this is very
> specific situation. I think that you will encounter this situation at
> the same probability that you will encounter any other kind of
> bottleneck (which is about 1/3 of the cases). However, as your example
> describe, to move out of the "CPU as a bottleneck" state into "I/O as
> bottleneck state" you have to scale up. Based on your example, your
> customer had to scale up from 4 way to 8 way.
> Now lets say that you paid the money and now you are I/O constrained.
> Ok, you buy more disks, change your RAID method, etc. Time passes and
> you are again constrained by the CPU... are you going to move to 16
> way?, 32 way? 64 way ?
> My point is that if you do not have CPU bottleneck now, you will be
> after you solve your current bottleneck. I.e. this is not VERY specific
> scenario but rather a common one.
> Tomer Meshorer
> Database Engineer
> Polyserve Inc (www.polyserve.com)
>
|||No. What I suggest is "Concurrent " execution, I.e. wall time speed up.
However, I cannot formulate and answer using babies and women while
staying
politically correct :-)
Tomer Meshorer
Database Engineer
Polyserve Inc (www.polyserve.com)
participate in load balancing and other two in fail over clustering.
My question is how will the two DB servers that is used for load
balancing synchronize?
Hi
To synchronize the data between the 2 clusters, you need to use something
like replication. Clustering is not a load-balancing technology.
Regards
Mike Epprecht, Microsoft SQL Server MVP
Zurich, Switzerland
IM: mike@.epprecht.net
MVP Program: http://www.microsoft.com/mvp
Blog: http://www.msmvps.com/epprecht/
<mukesh.thiru@.gmail.com> wrote in message
news:1135764099.960391.227140@.g43g2000cwa.googlegr oups.com...
>I have an architecture which has 4 DB servers on the whole; two will
> participate in load balancing and other two in fail over clustering.
> My question is how will the two DB servers that is used for load
> balancing synchronize?
>
|||Clustering is also not compatible with Windows Load Balancing, so you would
need to use a hardware based load balancer.
Mike
Mentor
Solid Quality Learning
http://www.solidqualitylearning.com
"Mike Epprecht (SQL MVP)" <mike@.epprecht.net> wrote in message
news:eQ%23n4Y6CGHA.620@.TK2MSFTNGP10.phx.gbl...
> Hi
> To synchronize the data between the 2 clusters, you need to use something
> like replication. Clustering is not a load-balancing technology.
> Regards
> --
> Mike Epprecht, Microsoft SQL Server MVP
> Zurich, Switzerland
> IM: mike@.epprecht.net
> MVP Program: http://www.microsoft.com/mvp
> Blog: http://www.msmvps.com/epprecht/
> <mukesh.thiru@.gmail.com> wrote in message
> news:1135764099.960391.227140@.g43g2000cwa.googlegr oups.com...
>
|||mukesh.thiru@.gmail.com wrote:
> I have an architecture which has 4 DB servers on the whole; two will
> participate in load balancing and other two in fail over clustering.
> My question is how will the two DB servers that is used for load
> balancing synchronize?
If you are using SQL server 2005, you might want to consider using read
only shared databases for your reporting . Using a shared file system,
all the nodes can see the same database files which are read only, and
mount the database as read only. You could run reporting quaries
against up to 16 nodes archiving close to 16X speedup.
If your update cycle and reporting cycle do not overlap, switching
between the read write to read only (and back) takes about 20 seconds.
Note that no replication is needed and no storage space is wasted.
Tomer Meshorer
Database Engineer
Polyserve Inc (http://www.polyserve.com)
Email:tmeshorer@.polyserve.com
|||I'd very strongly disagree with that assessment. It isn't a shared file
system, it requires a SAN.
Additionally, just because you point 2 SQL Server instances at the same
database does not mean you get 2X performance improvement, just like
pointing 16 instances at a single database will not give you 16X
performance. You could in fact see your performance degrade. The only way
that your performance would improve is if you have queries running on each
instance that are not competing with each other for disk throughput. If
there is disk contention, your performance will degrade since a single set
of disks are now servicing requests from multiple instances at the same
time.
Read only, shared databases is a nice new feature, but any performance
benefit you may get is very highly dependent upon your query patterns, the
volume of data in your active working set, and the amount of physical data
you can cache in memory on each instance.
Mike
Mentor
Solid Quality Learning
http://www.solidqualitylearning.com
"tmeshorer" <tmeshorer@.polyserve.com> wrote in message
news:1135935720.275938.306680@.g43g2000cwa.googlegr oups.com...
> mukesh.thiru@.gmail.com wrote:
> If you are using SQL server 2005, you might want to consider using read
> only shared databases for your reporting . Using a shared file system,
> all the nodes can see the same database files which are read only, and
> mount the database as read only. You could run reporting quaries
> against up to 16 nodes archiving close to 16X speedup.
> If your update cycle and reporting cycle do not overlap, switching
> between the read write to read only (and back) takes about 20 seconds.
> Note that no replication is needed and no storage space is wasted.
>
> Tomer Meshorer
> Database Engineer
> Polyserve Inc (http://www.polyserve.com)
> Email:tmeshorer@.polyserve.com
>
|||Your scenario is correct on a specific context. Vis you have an I/O
bottleneck
However, this is customer specific.
In cases where you have a CPU bottleneck (usually occurs in DW type of
quaries,which tends to do sequential I/O) the suggested solution will
scale close to linear(at least as observed in internal benchmarks where
I/O was not a bottleneck)
What I am suggesting, and specifically in the context of reporting, is
an alternative way for scaling SQL server.
Currently the only way to scale is up, which become very costly when
you go to the 8,16,32 CPU boxes.
In addition, you cannot scale economically since you move in increments
of 2^n each time you want to decrease your reporting time (and move
exponentially in price)
Note that in both scenarios (scale up to big SMP, or scale out on a
shared file system cluster) if you have an I/O bottleneck you reach the
state as described in your email.
However scaling out is more economically sound (assuming that you do
not have I/O bottle neck) since:
1. It is much cheaper
2. Pay as you go
3. No storage duplication
4. No changes to the database configuration(I.e. no need for
replication).
Best,
Tomer Meshorer
Database Engineer
Polyserve Inc
|||My only point was that you stated that you would get a 16X performance
improvement by using this when you deploy on 16 machines. That statement is
100% false and we both know it. You get ZERO performance improvement. Not
a single, solitary second of improved performance. A query running on the
operational system and taking 10 seconds is going to take 10 seconds to
complete if you are using a scalable shared database model.
Do not mix improved performance with the ability to execute more read only
queries per unit time. It is NOT the same thing and is an incredibly
misleading statement. It is misleading statements like that which lead to
people implementing things and then saying that "SQL Server can't scale" or
"SQL Server can't perform", because they are trying to make a feature do
something that it was not designed to do and never had the possibility of
doing in the first place.
You CAN be increased throughput and you CAN get increased capacity by being
able to run many more of these 10 second queries simultaneously when they
are executed against multiple machines. The ONLY time that you get
increased throughput (more queries per second) or increased capacity (more
queries per second) is if you hae a VERY specific situation where memory or
CPU is the ONLY bottleneck. In those situations, you can get increased
throughput or capacity, BUT, you do NOT and never will get a linear
increase in query capacity. The amount of increased capacity is VERY highly
dependent upon your database design, the volume of data you are working
with, and query patterns.
You cited DW type of queries. There is no such thing. You either have
queries that perform sequential I/O or queries performing random I/O.
Giving it a fancy name doesn't change it, because you run both types of
queries in every operational system. So, let's look at a CPU contrained
query. We have system where we have 4 processors and 4 GB of RAM. Queries
executing against the 400GB of data within the database normally take 15
minutes to execute and pull as much as 3 GB of data. We have isolated the
issue to CPU bottlenecks and moved the system to an 8 processor machine,
thereby doubling the CPU capacity. We observed that the execution time of
this query drops to 11 minutes whereby, we found that memory became the
bottleneck in the system. We doubled the RAM to 8GB and found that the
query execution only dropped to 10 minutes. Both memory and CPU became
bottlenecks at that point, because we were all of a sudden executing more
queries per unit time and 3 simultaneously executing queries would saturate
our memory. This would cause pages to be ejected from the cache (incurring
additional processor overhead). More pages would have to be read from disk
(more processor overhead). We quickly found out that it if we increased the
processor or memory capacity, we would then shift the bottleneck to the I/O
subsystem.
Sure, you can craft testing matrices and benchmarks that will say that
scalable, shared databases will get you a nearly linear scaling. (VERY
important to note that the word is SCALING, NOT performance improvement.)
For every scenario that you come up which exhibits this behavior, I can
change it a hundred different ways to cause the performance to degrade
either by changing the query pattern, introducing a new query, changing the
indexing, increasing the data volume, changing query parameters, etc.
It is a nice technology that has a VERY specific use in an environment
experiencing VERY specific scalability issues. Unless you have done a full
analysis on the system to determine whether it is appropriate, implementing
this technology would be a very big mistake. It most definitely isn't going
to produce a linear scaling and there is a reason that Microsoft is not
giving any specific numbers. (The instant they publish numbers like you
have, there are going to be several thousand people handing evidence that
the findings are not correct.)
This very specific issue is what makes all of the Oracle marketing and sales
surrounding RAC a complete joke. Oracle RAC won't produce a linear scaling
for exactly the same reasons that Scalable Shared Databases won't produce a
linear scaling. If your query pattern, active data volumes, and bottlenecks
meet a VERY specific profile then these types of technologies will get you
increased SCALABILITY. (That does NOT mean you get improved performance.)
Mike
Mentor
Solid Quality Learning
http://www.solidqualitylearning.com
"tmeshorer" <tmeshorer@.polyserve.com> wrote in message
news:1136012125.047011.237960@.g47g2000cwa.googlegr oups.com...
> Your scenario is correct on a specific context. Vis you have an I/O
> bottleneck
> However, this is customer specific.
> In cases where you have a CPU bottleneck (usually occurs in DW type of
> quaries,which tends to do sequential I/O) the suggested solution will
> scale close to linear(at least as observed in internal benchmarks where
> I/O was not a bottleneck)
> What I am suggesting, and specifically in the context of reporting, is
> an alternative way for scaling SQL server.
> Currently the only way to scale is up, which become very costly when
> you go to the 8,16,32 CPU boxes.
> In addition, you cannot scale economically since you move in increments
> of 2^n each time you want to decrease your reporting time (and move
> exponentially in price)
> Note that in both scenarios (scale up to big SMP, or scale out on a
> shared file system cluster) if you have an I/O bottleneck you reach the
> state as described in your email.
> However scaling out is more economically sound (assuming that you do
> not have I/O bottle neck) since:
> 1. It is much cheaper
> 2. Pay as you go
> 3. No storage duplication
> 4. No changes to the database configuration(I.e. no need for
> replication).
> Best,
> Tomer Meshorer
> Database Engineer
> Polyserve Inc
>
|||I should have clarified my point. I am looking at the actual wall clock
time that it takes to run a reporting job. I also assume that you have
more queries than servers, which I think would be the case for a
typical reporting job.
So from a wall clock time perspective, you will be able to get close to
X#server execution time improvement. I.e. if you are CPU constrained, a
6 hour job with 6 queries on one server can be reduced to 1 hour job on
6 servers.
If you have an I/O bottleneck there are two cases. If you are storage
constrained that this is a natural case. Vis, it does not matter if you
have 4*2 way servers or one 8 way server. If you are internal I/O BUS
constrains (e.g. PCI Express) moving to 4*2 way might improve your
situation since now you have 4 internal buses in your disposal.
However, I do not agree with your other point that this is very
specific situation. I think that you will encounter this situation at
the same probability that you will encounter any other kind of
bottleneck (which is about 1/3 of the cases). However, as your example
describe, to move out of the "CPU as a bottleneck" state into "I/O as
bottleneck state" you have to scale up. Based on your example, your
customer had to scale up from 4 way to 8 way.
Now lets say that you paid the money and now you are I/O constrained.
Ok, you buy more disks, change your RAID method, etc. Time passes and
you are again constrained by the CPU... are you going to move to 16
way?, 32 way? 64 way ?
My point is that if you do not have CPU bottleneck now, you will be
after you solve your current bottleneck. I.e. this is not VERY specific
scenario but rather a common one.
Tomer Meshorer
Database Engineer
Polyserve Inc (www.polyserve.com)
|||So that means that I can put 9 women to work and produce a baby in 1 month?
Mike
Mentor
Solid Quality Learning
http://www.solidqualitylearning.com
"tmeshorer" <tmeshorer@.polyserve.com> wrote in message
news:1136319106.372823.46070@.o13g2000cwo.googlegro ups.com...
>I should have clarified my point. I am looking at the actual wall clock
> time that it takes to run a reporting job. I also assume that you have
> more queries than servers, which I think would be the case for a
> typical reporting job.
> So from a wall clock time perspective, you will be able to get close to
> X#server execution time improvement. I.e. if you are CPU constrained, a
> 6 hour job with 6 queries on one server can be reduced to 1 hour job on
> 6 servers.
> If you have an I/O bottleneck there are two cases. If you are storage
> constrained that this is a natural case. Vis, it does not matter if you
> have 4*2 way servers or one 8 way server. If you are internal I/O BUS
> constrains (e.g. PCI Express) moving to 4*2 way might improve your
> situation since now you have 4 internal buses in your disposal.
> However, I do not agree with your other point that this is very
> specific situation. I think that you will encounter this situation at
> the same probability that you will encounter any other kind of
> bottleneck (which is about 1/3 of the cases). However, as your example
> describe, to move out of the "CPU as a bottleneck" state into "I/O as
> bottleneck state" you have to scale up. Based on your example, your
> customer had to scale up from 4 way to 8 way.
> Now lets say that you paid the money and now you are I/O constrained.
> Ok, you buy more disks, change your RAID method, etc. Time passes and
> you are again constrained by the CPU... are you going to move to 16
> way?, 32 way? 64 way ?
> My point is that if you do not have CPU bottleneck now, you will be
> after you solve your current bottleneck. I.e. this is not VERY specific
> scenario but rather a common one.
> Tomer Meshorer
> Database Engineer
> Polyserve Inc (www.polyserve.com)
>
|||No. What I suggest is "Concurrent " execution, I.e. wall time speed up.
However, I cannot formulate and answer using babies and women while
staying
politically correct :-)
Tomer Meshorer
Database Engineer
Polyserve Inc (www.polyserve.com)
Cluster Switch Hardware
New to clustering and wanted to know if anyone had any hardware suggestions
with regard to the dedicated switch for the heartbeat network.
Initially this will be a 2 node Active/Active cluster hosting SQL Server
connected to a SAN. I do expect to add at least another set of clustered
hosts or two before year's end.
We typically use all Cisco equipment.
Suggestions?
"john d" <johnd@.discussions.microsoft.com> wrote in message
news:D9FCFA8C-13FB-4A7C-9E88-9573BEF9C047@.microsoft.com...
> New to clustering and wanted to know if anyone had any hardware
> suggestions
> with regard to the dedicated switch for the heartbeat network.
Anything will work for you. I, personally, prefer dumb hubs. That way Bucky
can misconfigure it and hose up my heartbeat network.
Russ Kaufmann
MVP - Windows Server - Clustering
ClusterHelp.com, a Microsoft Certified Gold Partner
Web http://www.clusterhelp.com
Blog http://msmvps.com/clusterhelp
The next ClusterHelp classes are:
Denver starting Feb 12th
NYC starting Feb 19th
|||I agree with Russ. The bandwith requirements for the heartbeat are really
low. I wish I could get some old co-ax NICS and just run a single cable
instead of an active device.
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP
"Russ Kaufmann [MVP]" <russ@.clusterhelp.com> wrote in message
news:erF4iczOHHA.4280@.TK2MSFTNGP02.phx.gbl...
> "john d" <johnd@.discussions.microsoft.com> wrote in message
> news:D9FCFA8C-13FB-4A7C-9E88-9573BEF9C047@.microsoft.com...
> Anything will work for you. I, personally, prefer dumb hubs. That way
> Bucky can misconfigure it and hose up my heartbeat network.
>
> --
> Russ Kaufmann
> MVP - Windows Server - Clustering
> ClusterHelp.com, a Microsoft Certified Gold Partner
> Web http://www.clusterhelp.com
> Blog http://msmvps.com/clusterhelp
> The next ClusterHelp classes are:
> Denver starting Feb 12th
> NYC starting Feb 19th
>
|||Geoff/Russ - Thanks for your input. Glad I didn't already go out and buy a
switch.
"Geoff N. Hiten" wrote:
> I agree with Russ. The bandwith requirements for the heartbeat are really
> low. I wish I could get some old co-ax NICS and just run a single cable
> instead of an active device.
> --
> Geoff N. Hiten
> Senior Database Administrator
> Microsoft SQL Server MVP
>
> "Russ Kaufmann [MVP]" <russ@.clusterhelp.com> wrote in message
> news:erF4iczOHHA.4280@.TK2MSFTNGP02.phx.gbl...
>
|||For a 2-node system, we typically use just a network cross-over cable. If
you expand this cluster for additional nodes, the heartbeat uses multi-cast
to communicate (Win2K3); so, you will definitely want to isolate this
network from the others, either by constructing a physical interconnect
between the servers or through a larger corporate switch interconnect and
the use of VLANs.
Regardless, unless you also duplicate the hardware, this network will also
represent a Single Point of Failure. If you are unfamiliar with this term,
you should learn it. It is the single most important concept with regards
to Highly Available systems.
In this case, you must provide multiple network paths between the servers.
In addition to the dedicated heartbeat network, we typically allow the
public network to also transport cluster heartbeat communications. In
Cluster Administrator, on the cluster object, you can specify the network
priority for cluster communications, but also include the secondary network
as a fail-safe.
Sincerely,
Anthony Thomas
"john d" <johnd@.discussions.microsoft.com> wrote in message
news:DC09A8BC-4126-4002-B526-401B90D2A1DB@.microsoft.com...
> Geoff/Russ - Thanks for your input. Glad I didn't already go out and buy
a[vbcol=seagreen]
> switch.
>
> "Geoff N. Hiten" wrote:
really[vbcol=seagreen]
with regard to the dedicated switch for the heartbeat network.
Initially this will be a 2 node Active/Active cluster hosting SQL Server
connected to a SAN. I do expect to add at least another set of clustered
hosts or two before year's end.
We typically use all Cisco equipment.
Suggestions?
"john d" <johnd@.discussions.microsoft.com> wrote in message
news:D9FCFA8C-13FB-4A7C-9E88-9573BEF9C047@.microsoft.com...
> New to clustering and wanted to know if anyone had any hardware
> suggestions
> with regard to the dedicated switch for the heartbeat network.
Anything will work for you. I, personally, prefer dumb hubs. That way Bucky
can misconfigure it and hose up my heartbeat network.
Russ Kaufmann
MVP - Windows Server - Clustering
ClusterHelp.com, a Microsoft Certified Gold Partner
Web http://www.clusterhelp.com
Blog http://msmvps.com/clusterhelp
The next ClusterHelp classes are:
Denver starting Feb 12th
NYC starting Feb 19th
|||I agree with Russ. The bandwith requirements for the heartbeat are really
low. I wish I could get some old co-ax NICS and just run a single cable
instead of an active device.
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP
"Russ Kaufmann [MVP]" <russ@.clusterhelp.com> wrote in message
news:erF4iczOHHA.4280@.TK2MSFTNGP02.phx.gbl...
> "john d" <johnd@.discussions.microsoft.com> wrote in message
> news:D9FCFA8C-13FB-4A7C-9E88-9573BEF9C047@.microsoft.com...
> Anything will work for you. I, personally, prefer dumb hubs. That way
> Bucky can misconfigure it and hose up my heartbeat network.
>
> --
> Russ Kaufmann
> MVP - Windows Server - Clustering
> ClusterHelp.com, a Microsoft Certified Gold Partner
> Web http://www.clusterhelp.com
> Blog http://msmvps.com/clusterhelp
> The next ClusterHelp classes are:
> Denver starting Feb 12th
> NYC starting Feb 19th
>
|||Geoff/Russ - Thanks for your input. Glad I didn't already go out and buy a
switch.
"Geoff N. Hiten" wrote:
> I agree with Russ. The bandwith requirements for the heartbeat are really
> low. I wish I could get some old co-ax NICS and just run a single cable
> instead of an active device.
> --
> Geoff N. Hiten
> Senior Database Administrator
> Microsoft SQL Server MVP
>
> "Russ Kaufmann [MVP]" <russ@.clusterhelp.com> wrote in message
> news:erF4iczOHHA.4280@.TK2MSFTNGP02.phx.gbl...
>
|||For a 2-node system, we typically use just a network cross-over cable. If
you expand this cluster for additional nodes, the heartbeat uses multi-cast
to communicate (Win2K3); so, you will definitely want to isolate this
network from the others, either by constructing a physical interconnect
between the servers or through a larger corporate switch interconnect and
the use of VLANs.
Regardless, unless you also duplicate the hardware, this network will also
represent a Single Point of Failure. If you are unfamiliar with this term,
you should learn it. It is the single most important concept with regards
to Highly Available systems.
In this case, you must provide multiple network paths between the servers.
In addition to the dedicated heartbeat network, we typically allow the
public network to also transport cluster heartbeat communications. In
Cluster Administrator, on the cluster object, you can specify the network
priority for cluster communications, but also include the secondary network
as a fail-safe.
Sincerely,
Anthony Thomas
"john d" <johnd@.discussions.microsoft.com> wrote in message
news:DC09A8BC-4126-4002-B526-401B90D2A1DB@.microsoft.com...
> Geoff/Russ - Thanks for your input. Glad I didn't already go out and buy
a[vbcol=seagreen]
> switch.
>
> "Geoff N. Hiten" wrote:
really[vbcol=seagreen]
cluster SSAS and SQL
Hi,
We have a default clusterd SQL server 2005 with sp2 installed in one cluster
group and then I have installed a default SSAS clustered instance in another
cluster group and it works fine (can failover and user can connect). However,
when I applying the sp2 to SSAS instance (with SSAS checked only) and it
failed; it not noly just failed but also take down my default clustered sql
server! We have to call MS to get our server up. It seems that, somehow,
registry for default sql server got messed up by SSAS sp2. My question are:
1.Is it possible or supported to install a default SQL server and a default
Analysis server on a cluster (each one in its own cluster group) with sp2;
2.why applying sp2 on a default AD instance in one cluster group affect a
default sql instance in another cluster group.
thanks much in advance.
Zack.
The answer is that SQL services share some common components. Plus, the
installer has some generic setup logic. Finally, are you using the same
service account for both services? If so, then permissions can get hosed
during a broken update.
Geoff N. Hiten
Senior SQL Infrastructure Consultant
Microsoft SQL Server MVP
"dp" <dp@.discussions.microsoft.com> wrote in message
news:90BAC728-916F-4497-B293-55920962EDE8@.microsoft.com...
> Hi,
> We have a default clusterd SQL server 2005 with sp2 installed in one
> cluster
> group and then I have installed a default SSAS clustered instance in
> another
> cluster group and it works fine (can failover and user can connect).
> However,
> when I applying the sp2 to SSAS instance (with SSAS checked only) and it
> failed; it not noly just failed but also take down my default clustered
> sql
> server! We have to call MS to get our server up. It seems that, somehow,
> registry for default sql server got messed up by SSAS sp2. My question
> are:
> 1. Is it possible or supported to install a default SQL server and a
> default
> Analysis server on a cluster (each one in its own cluster group) with sp2;
> 2. why applying sp2 on a default AD instance in one cluster group affect a
> default sql instance in another cluster group.
> thanks much in advance.
> Zack.
>
|||Zack, were you ever able to upgrade your SSAS in the cluster?
We are having the same or similar issues. We upgraded our cluster from SP1
to SP2. The SQL Server components upgraded to SP2, but the SSAS failed and
is still at SP1. After the upgrade one of our clustered instances did not
come back online. Error logs read something like, "...Cluster IP/Network
resource already in use..."
Well, I called MS because our production SQL 2005 cluster was hosed. The
issue, reg key "HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microso ft SQL
Server\MSSQL.1\Cluster\ClusterName" had the incorrect instance name for the
MSSQL.1 instance (in our case). We had to run some cluster commands to
remove the MSCS checkpoint, changed the key value to the correct instance
name, then reapply MSCS checkpoint.
This got our SQL Server instance back online. But we still can't update
SSAS to SP2. Maybe the two issues are not related, but I have a gut feeling
they are. The incorrect 'ClusterName' value happened to be the instance name
for our SSAS.
I still have the MS support ticket open trying to determine if the cluster
registry keys are in the correct state on both nodes.
Anyone else having issues with failover cluster anaylsis server upgrade to
SP2?
"dp" wrote:
> Hi,
> We have a default clusterd SQL server 2005 with sp2 installed in one cluster
> group and then I have installed a default SSAS clustered instance in another
> cluster group and it works fine (can failover and user can connect). However,
> when I applying the sp2 to SSAS instance (with SSAS checked only) and it
> failed; it not noly just failed but also take down my default clustered sql
> server! We have to call MS to get our server up. It seems that, somehow,
> registry for default sql server got messed up by SSAS sp2. My question are:
> 1.Is it possible or supported to install a default SQL server and a default
> Analysis server on a cluster (each one in its own cluster group) with sp2;
> 2.why applying sp2 on a default AD instance in one cluster group affect a
> default sql instance in another cluster group.
> thanks much in advance.
> Zack.
>
We have a default clusterd SQL server 2005 with sp2 installed in one cluster
group and then I have installed a default SSAS clustered instance in another
cluster group and it works fine (can failover and user can connect). However,
when I applying the sp2 to SSAS instance (with SSAS checked only) and it
failed; it not noly just failed but also take down my default clustered sql
server! We have to call MS to get our server up. It seems that, somehow,
registry for default sql server got messed up by SSAS sp2. My question are:
1.Is it possible or supported to install a default SQL server and a default
Analysis server on a cluster (each one in its own cluster group) with sp2;
2.why applying sp2 on a default AD instance in one cluster group affect a
default sql instance in another cluster group.
thanks much in advance.
Zack.
The answer is that SQL services share some common components. Plus, the
installer has some generic setup logic. Finally, are you using the same
service account for both services? If so, then permissions can get hosed
during a broken update.
Geoff N. Hiten
Senior SQL Infrastructure Consultant
Microsoft SQL Server MVP
"dp" <dp@.discussions.microsoft.com> wrote in message
news:90BAC728-916F-4497-B293-55920962EDE8@.microsoft.com...
> Hi,
> We have a default clusterd SQL server 2005 with sp2 installed in one
> cluster
> group and then I have installed a default SSAS clustered instance in
> another
> cluster group and it works fine (can failover and user can connect).
> However,
> when I applying the sp2 to SSAS instance (with SSAS checked only) and it
> failed; it not noly just failed but also take down my default clustered
> sql
> server! We have to call MS to get our server up. It seems that, somehow,
> registry for default sql server got messed up by SSAS sp2. My question
> are:
> 1. Is it possible or supported to install a default SQL server and a
> default
> Analysis server on a cluster (each one in its own cluster group) with sp2;
> 2. why applying sp2 on a default AD instance in one cluster group affect a
> default sql instance in another cluster group.
> thanks much in advance.
> Zack.
>
|||Zack, were you ever able to upgrade your SSAS in the cluster?
We are having the same or similar issues. We upgraded our cluster from SP1
to SP2. The SQL Server components upgraded to SP2, but the SSAS failed and
is still at SP1. After the upgrade one of our clustered instances did not
come back online. Error logs read something like, "...Cluster IP/Network
resource already in use..."
Well, I called MS because our production SQL 2005 cluster was hosed. The
issue, reg key "HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microso ft SQL
Server\MSSQL.1\Cluster\ClusterName" had the incorrect instance name for the
MSSQL.1 instance (in our case). We had to run some cluster commands to
remove the MSCS checkpoint, changed the key value to the correct instance
name, then reapply MSCS checkpoint.
This got our SQL Server instance back online. But we still can't update
SSAS to SP2. Maybe the two issues are not related, but I have a gut feeling
they are. The incorrect 'ClusterName' value happened to be the instance name
for our SSAS.
I still have the MS support ticket open trying to determine if the cluster
registry keys are in the correct state on both nodes.
Anyone else having issues with failover cluster anaylsis server upgrade to
SP2?
"dp" wrote:
> Hi,
> We have a default clusterd SQL server 2005 with sp2 installed in one cluster
> group and then I have installed a default SSAS clustered instance in another
> cluster group and it works fine (can failover and user can connect). However,
> when I applying the sp2 to SSAS instance (with SSAS checked only) and it
> failed; it not noly just failed but also take down my default clustered sql
> server! We have to call MS to get our server up. It seems that, somehow,
> registry for default sql server got messed up by SSAS sp2. My question are:
> 1.Is it possible or supported to install a default SQL server and a default
> Analysis server on a cluster (each one in its own cluster group) with sp2;
> 2.why applying sp2 on a default AD instance in one cluster group affect a
> default sql instance in another cluster group.
> thanks much in advance.
> Zack.
>
cluster sql2000 & DTC
on the passive node i tried to set range port for DTC
I tried using dcomcnfg, but when i reboot the node cannot join the cluster.
Why ?
thanks in advance.
Regards,
Dario
Review the NT event logs and cluster.log to get more details on why the node failed to join the cluster.
Best Regards,
Uttam Parui
Microsoft Corporation
This posting is provided "AS IS" with no warranties, and confers no rights.
Are you secure? For information about the Strategic Technology Protection Program and to order your FREE Security Tool Kit, please visit http://www.microsoft.com/security.
Microsoft highly recommends that users with Internet access update their Microsoft software to better protect against viruses and security vulnerabilities. The easiest way to do this is to visit the following websites:
http://www.microsoft.com/protect
http://www.microsoft.com/security/guidance/default.mspx
|||DTC by deault will use any port ranges.
a registry hack can be done to limit it to certain range of ports.
"Dario Sala" wrote:
> on the passive node i tried to set range port for DTC
> I tried using dcomcnfg, but when i reboot the node cannot join the cluster.
> Why ?
> thanks in advance.
> Regards,
> Dario
>
>
I tried using dcomcnfg, but when i reboot the node cannot join the cluster.
Why ?
thanks in advance.
Regards,
Dario
Review the NT event logs and cluster.log to get more details on why the node failed to join the cluster.
Best Regards,
Uttam Parui
Microsoft Corporation
This posting is provided "AS IS" with no warranties, and confers no rights.
Are you secure? For information about the Strategic Technology Protection Program and to order your FREE Security Tool Kit, please visit http://www.microsoft.com/security.
Microsoft highly recommends that users with Internet access update their Microsoft software to better protect against viruses and security vulnerabilities. The easiest way to do this is to visit the following websites:
http://www.microsoft.com/protect
http://www.microsoft.com/security/guidance/default.mspx
|||DTC by deault will use any port ranges.
a registry hack can be done to limit it to certain range of ports.
"Dario Sala" wrote:
> on the passive node i tried to set range port for DTC
> I tried using dcomcnfg, but when i reboot the node cannot join the cluster.
> Why ?
> thanks in advance.
> Regards,
> Dario
>
>
cluster sql server 2000
Hi,
I need manuals, tecnical papres, instalation guide, etc about
how to install one cluster sql server 2000 64 bits with
windows server 2003.
It`s posible a need one document step by step about this
Thanks,
RaulGZ."raulgz" <ragaza@.ozu.es> wrote in message
news:9b551742.0402270130.5c1c5655@.posting.google.c om...
> Hi,
> I need manuals, tecnical papres, instalation guide, etc about
> how to install one cluster sql server 2000 64 bits with
> windows server 2003.
> It`s posible a need one document step by step about this
>
> Thanks,
> RaulGZ.
http://www.microsoft.com/technet/tr...at/failclus.asp
Cluster SQL Active-Active
My friends:
I need document information about the installation process of SQL in Cluster
with two instances, both active, one in each Server. In summary I want to
install a Cluster of SQL Active-Active, for take greater benefit of both
Cluster nodes.
Even though I have looked for in BOL (follow the recommendations of Maxi), I
had investigate and knock the same trouble, there are only information about
the Active-Pasive installation, and I'll wish a procedure or document about
Active-Active setup and installation.
If someone knows where I can find the information, or somebody understand
about Active-Active installation, please give it me know. Thanks a lot.
Regards,
Carlo Sorrel
Hello Carlo,
Once you have installed the first instance of virtual SQL Server 2000 then the steps for installing the second instance of virtual SQL Server 2000 is exactly the same. The second virtual instance of SQL Server 2000
will need a seperate shared drive (you cannot use the shared drive that you used for the first instance), unique SQL IP Address (again you cannot use the one you used for your windows cluster ip or the first sql
server ip) and unique SQL network name. Also, if the first virtual instance is a default instance then the second instance can ONLY be a named instance. You CANNOT have two default virtual SQL Server 2000
instance on the same cluster. To have both the instances similar, many customers install both instances of virtual SQL Server 2000 as named instances.
Since you have multiple instances (2 in your case), you will need to ensure that all server nodes have the same resources (processor, memory) and it is enough to handle the instances that could potentially fail to
that node. Another important consideration is to cap memory usage of the instance of SQL Server 2000 with max server memory. Especially if AWE memory is enabled, max server memory must be set in a
multiple-instance cluster to prevent starving the server node.
If you have a Windows Server 2003 EE cluster then you may find the following webcast useful
TechNet Support WebCast: How to install a Microsoft SQL Server 2000 virtual server on a Microsoft Windows Server 2003 cluster
Discusses how to install Microsoft SQL Server 2000 clustering on Microsoft Windows Server 2003 and differences between Microsoft Windows 2000 Server and Windows Server 2003.
http://support.microsoft.com/default...b;en-us;888121
For Win2K Adv Server cluster, you may find the following whitepaper useful
SQL Server 2000 Failover Clustering
http://www.microsoft.com/technet/pro.../failclus.mspx
Review the above and feel free to post any other qs that you may have.
Additional Information
=======================
Here is some other related links that you may find useful
INF: Clustered SQL Server Do's, Don'ts, and Basic Warnings
http://support.microsoft.com/?kbid=254321
Introduction to Microsoft SQL Server 2000 Clustering
http://support.microsoft.com/default...lurb051001.asp
Microsoft SQL Server 2000 Virtual Server: Things You Should Know
http://support.microsoft.com/default...lurb032602.asp
Microsoft SQL Server 2000 Virtual Server Basic Setup, Maintenance, and Service Pack http://support.microsoft.com/default...lurb061002.asp
Microsoft SQL Server 2000 Failover Clustering Disaster Recovery Procedures
http://support.microsoft.com/default...lurb101802.asp
Troubleshooting SQL 2000 Virtual Server and Service Pack Setups for Failover Clustering
http://support.microsoft.com/default...lurb020703.asp
HTH,
Best Regards,
Uttam Parui
Microsoft Corporation
This posting is provided "AS IS" with no warranties, and confers no rights.
Are you secure? For information about the Strategic Technology Protection Program and to order your FREE Security Tool Kit, please visit http://www.microsoft.com/security.
Microsoft highly recommends that users with Internet access update their Microsoft software to better protect against viruses and security vulnerabilities. The easiest way to do this is to visit the following websites:
http://www.microsoft.com/protect
http://www.microsoft.com/security/guidance/default.mspx
|||Hi
Just be aware, Active-Active does not mean load balancing. It means 2
instances of different databases, each running by default on one of the
nodes of a cluster.
The references Uttam has supplied are great and will help you along.
Regards
Mike Epprecht, Microsoft SQL Server MVP
Zurich, Switzerland
IM: mike@.epprecht.net
MVP Program: http://www.microsoft.com/mvp
Blog: http://www.msmvps.com/epprecht/
"Carlo Sorrel" <csorrel@.hotmail.com> wrote in message
news:#nPFJo7AFHA.3576@.TK2MSFTNGP11.phx.gbl...
> My friends:
>
> I need document information about the installation process of SQL in
Cluster
> with two instances, both active, one in each Server. In summary I want to
> install a Cluster of SQL Active-Active, for take greater benefit of both
> Cluster nodes.
>
> Even though I have looked for in BOL (follow the recommendations of Maxi),
I
> had investigate and knock the same trouble, there are only information
about
> the Active-Pasive installation, and I'll wish a procedure or document
about
> Active-Active setup and installation.
>
> If someone knows where I can find the information, or somebody understand
> about Active-Active installation, please give it me know. Thanks a lot.
>
> Regards,
>
> Carlo Sorrel
>
|||first, sorry my english..., apears this error during instalattion the second instance on Cluster Windows 2003.
The description for Event ID ( 17052 ) in Source ( MSSQL$LASCAR ) cannot be found. The local computer may not have the necessary registry information or message DLL files to display messages from a remote computer. You may be able to use the /AUXSOURCE= flag to retrieve this description; see Help and Support for details. The following information is part of the event: [sqsrvres] checkODBCConnectError: sqlstate = 01000; native error = 35; message = [Microsoft][ODBC SQL Server Driver][DBNETLIB]ConnectionOpen (Connect()).
Tahnk's.
Carlo Sorrel.
"Uttam Parui[MS]" <uttamkp@.online.microsoft.com> escribi en el mensaje news:uX597q8AFHA.1680@.cpmsftngxa10.phx.gbl...
> Hello Carlo,
> Once you have installed the first instance of virtual SQL Server 2000 then the steps for installing the second instance of virtual SQL Server 2000 is exactly the same. The second virtual instance of SQL Server 2000
> will need a seperate shared drive (you cannot use the shared drive that you used for the first instance), unique SQL IP Address (again you cannot use the one you used for your windows cluster ip or the first sql
> server ip) and unique SQL network name. Also, if the first virtual instance is a default instance then the second instance can ONLY be a named instance. You CANNOT have two default virtual SQL Server 2000
> instance on the same cluster. To have both the instances similar, many customers install both instances of virtual SQL Server 2000 as named instances.
> Since you have multiple instances (2 in your case), you will need to ensure that all server nodes have the same resources (processor, memory) and it is enough to handle the instances that could potentially fail to
> that node. Another important consideration is to cap memory usage of the instance of SQL Server 2000 with max server memory. Especially if AWE memory is enabled, max server memory must be set in a
> multiple-instance cluster to prevent starving the server node.
> If you have a Windows Server 2003 EE cluster then you may find the following webcast useful
> TechNet Support WebCast: How to install a Microsoft SQL Server 2000 virtual server on a Microsoft Windows Server 2003 cluster
> Discusses how to install Microsoft SQL Server 2000 clustering on Microsoft Windows Server 2003 and differences between Microsoft Windows 2000 Server and Windows Server 2003.
> http://support.microsoft.com/default...b;en-us;888121
>
> For Win2K Adv Server cluster, you may find the following whitepaper useful
> SQL Server 2000 Failover Clustering
> http://www.microsoft.com/technet/pro.../failclus.mspx
> Review the above and feel free to post any other qs that you may have.
> Additional Information
> =======================
> Here is some other related links that you may find useful
> INF: Clustered SQL Server Do's, Don'ts, and Basic Warnings
> http://support.microsoft.com/?kbid=254321
> Introduction to Microsoft SQL Server 2000 Clustering
> http://support.microsoft.com/default...lurb051001.asp
> Microsoft SQL Server 2000 Virtual Server: Things You Should Know
> http://support.microsoft.com/default...lurb032602.asp
>
> Microsoft SQL Server 2000 Virtual Server Basic Setup, Maintenance, and Service Pack http://support.microsoft.com/default...lurb061002.asp
> Microsoft SQL Server 2000 Failover Clustering Disaster Recovery Procedures
> http://support.microsoft.com/default...lurb101802.asp
> Troubleshooting SQL 2000 Virtual Server and Service Pack Setups for Failover Clustering
> http://support.microsoft.com/default...lurb020703.asp
>
> HTH,
> Best Regards,
> Uttam Parui
> Microsoft Corporation
> This posting is provided "AS IS" with no warranties, and confers no rights.
> Are you secure? For information about the Strategic Technology Protection Program and to order your FREE Security Tool Kit, please visit http://www.microsoft.com/security.
> Microsoft highly recommends that users with Internet access update their Microsoft software to better protect against viruses and security vulnerabilities. The easiest way to do this is to visit the following websites:
> http://www.microsoft.com/protect
> http://www.microsoft.com/security/guidance/default.mspx
>
I need document information about the installation process of SQL in Cluster
with two instances, both active, one in each Server. In summary I want to
install a Cluster of SQL Active-Active, for take greater benefit of both
Cluster nodes.
Even though I have looked for in BOL (follow the recommendations of Maxi), I
had investigate and knock the same trouble, there are only information about
the Active-Pasive installation, and I'll wish a procedure or document about
Active-Active setup and installation.
If someone knows where I can find the information, or somebody understand
about Active-Active installation, please give it me know. Thanks a lot.
Regards,
Carlo Sorrel
Hello Carlo,
Once you have installed the first instance of virtual SQL Server 2000 then the steps for installing the second instance of virtual SQL Server 2000 is exactly the same. The second virtual instance of SQL Server 2000
will need a seperate shared drive (you cannot use the shared drive that you used for the first instance), unique SQL IP Address (again you cannot use the one you used for your windows cluster ip or the first sql
server ip) and unique SQL network name. Also, if the first virtual instance is a default instance then the second instance can ONLY be a named instance. You CANNOT have two default virtual SQL Server 2000
instance on the same cluster. To have both the instances similar, many customers install both instances of virtual SQL Server 2000 as named instances.
Since you have multiple instances (2 in your case), you will need to ensure that all server nodes have the same resources (processor, memory) and it is enough to handle the instances that could potentially fail to
that node. Another important consideration is to cap memory usage of the instance of SQL Server 2000 with max server memory. Especially if AWE memory is enabled, max server memory must be set in a
multiple-instance cluster to prevent starving the server node.
If you have a Windows Server 2003 EE cluster then you may find the following webcast useful
TechNet Support WebCast: How to install a Microsoft SQL Server 2000 virtual server on a Microsoft Windows Server 2003 cluster
Discusses how to install Microsoft SQL Server 2000 clustering on Microsoft Windows Server 2003 and differences between Microsoft Windows 2000 Server and Windows Server 2003.
http://support.microsoft.com/default...b;en-us;888121
For Win2K Adv Server cluster, you may find the following whitepaper useful
SQL Server 2000 Failover Clustering
http://www.microsoft.com/technet/pro.../failclus.mspx
Review the above and feel free to post any other qs that you may have.
Additional Information
=======================
Here is some other related links that you may find useful
INF: Clustered SQL Server Do's, Don'ts, and Basic Warnings
http://support.microsoft.com/?kbid=254321
Introduction to Microsoft SQL Server 2000 Clustering
http://support.microsoft.com/default...lurb051001.asp
Microsoft SQL Server 2000 Virtual Server: Things You Should Know
http://support.microsoft.com/default...lurb032602.asp
Microsoft SQL Server 2000 Virtual Server Basic Setup, Maintenance, and Service Pack http://support.microsoft.com/default...lurb061002.asp
Microsoft SQL Server 2000 Failover Clustering Disaster Recovery Procedures
http://support.microsoft.com/default...lurb101802.asp
Troubleshooting SQL 2000 Virtual Server and Service Pack Setups for Failover Clustering
http://support.microsoft.com/default...lurb020703.asp
HTH,
Best Regards,
Uttam Parui
Microsoft Corporation
This posting is provided "AS IS" with no warranties, and confers no rights.
Are you secure? For information about the Strategic Technology Protection Program and to order your FREE Security Tool Kit, please visit http://www.microsoft.com/security.
Microsoft highly recommends that users with Internet access update their Microsoft software to better protect against viruses and security vulnerabilities. The easiest way to do this is to visit the following websites:
http://www.microsoft.com/protect
http://www.microsoft.com/security/guidance/default.mspx
|||Hi
Just be aware, Active-Active does not mean load balancing. It means 2
instances of different databases, each running by default on one of the
nodes of a cluster.
The references Uttam has supplied are great and will help you along.
Regards
Mike Epprecht, Microsoft SQL Server MVP
Zurich, Switzerland
IM: mike@.epprecht.net
MVP Program: http://www.microsoft.com/mvp
Blog: http://www.msmvps.com/epprecht/
"Carlo Sorrel" <csorrel@.hotmail.com> wrote in message
news:#nPFJo7AFHA.3576@.TK2MSFTNGP11.phx.gbl...
> My friends:
>
> I need document information about the installation process of SQL in
Cluster
> with two instances, both active, one in each Server. In summary I want to
> install a Cluster of SQL Active-Active, for take greater benefit of both
> Cluster nodes.
>
> Even though I have looked for in BOL (follow the recommendations of Maxi),
I
> had investigate and knock the same trouble, there are only information
about
> the Active-Pasive installation, and I'll wish a procedure or document
about
> Active-Active setup and installation.
>
> If someone knows where I can find the information, or somebody understand
> about Active-Active installation, please give it me know. Thanks a lot.
>
> Regards,
>
> Carlo Sorrel
>
|||first, sorry my english..., apears this error during instalattion the second instance on Cluster Windows 2003.
The description for Event ID ( 17052 ) in Source ( MSSQL$LASCAR ) cannot be found. The local computer may not have the necessary registry information or message DLL files to display messages from a remote computer. You may be able to use the /AUXSOURCE= flag to retrieve this description; see Help and Support for details. The following information is part of the event: [sqsrvres] checkODBCConnectError: sqlstate = 01000; native error = 35; message = [Microsoft][ODBC SQL Server Driver][DBNETLIB]ConnectionOpen (Connect()).
Tahnk's.
Carlo Sorrel.
"Uttam Parui[MS]" <uttamkp@.online.microsoft.com> escribi en el mensaje news:uX597q8AFHA.1680@.cpmsftngxa10.phx.gbl...
> Hello Carlo,
> Once you have installed the first instance of virtual SQL Server 2000 then the steps for installing the second instance of virtual SQL Server 2000 is exactly the same. The second virtual instance of SQL Server 2000
> will need a seperate shared drive (you cannot use the shared drive that you used for the first instance), unique SQL IP Address (again you cannot use the one you used for your windows cluster ip or the first sql
> server ip) and unique SQL network name. Also, if the first virtual instance is a default instance then the second instance can ONLY be a named instance. You CANNOT have two default virtual SQL Server 2000
> instance on the same cluster. To have both the instances similar, many customers install both instances of virtual SQL Server 2000 as named instances.
> Since you have multiple instances (2 in your case), you will need to ensure that all server nodes have the same resources (processor, memory) and it is enough to handle the instances that could potentially fail to
> that node. Another important consideration is to cap memory usage of the instance of SQL Server 2000 with max server memory. Especially if AWE memory is enabled, max server memory must be set in a
> multiple-instance cluster to prevent starving the server node.
> If you have a Windows Server 2003 EE cluster then you may find the following webcast useful
> TechNet Support WebCast: How to install a Microsoft SQL Server 2000 virtual server on a Microsoft Windows Server 2003 cluster
> Discusses how to install Microsoft SQL Server 2000 clustering on Microsoft Windows Server 2003 and differences between Microsoft Windows 2000 Server and Windows Server 2003.
> http://support.microsoft.com/default...b;en-us;888121
>
> For Win2K Adv Server cluster, you may find the following whitepaper useful
> SQL Server 2000 Failover Clustering
> http://www.microsoft.com/technet/pro.../failclus.mspx
> Review the above and feel free to post any other qs that you may have.
> Additional Information
> =======================
> Here is some other related links that you may find useful
> INF: Clustered SQL Server Do's, Don'ts, and Basic Warnings
> http://support.microsoft.com/?kbid=254321
> Introduction to Microsoft SQL Server 2000 Clustering
> http://support.microsoft.com/default...lurb051001.asp
> Microsoft SQL Server 2000 Virtual Server: Things You Should Know
> http://support.microsoft.com/default...lurb032602.asp
>
> Microsoft SQL Server 2000 Virtual Server Basic Setup, Maintenance, and Service Pack http://support.microsoft.com/default...lurb061002.asp
> Microsoft SQL Server 2000 Failover Clustering Disaster Recovery Procedures
> http://support.microsoft.com/default...lurb101802.asp
> Troubleshooting SQL 2000 Virtual Server and Service Pack Setups for Failover Clustering
> http://support.microsoft.com/default...lurb020703.asp
>
> HTH,
> Best Regards,
> Uttam Parui
> Microsoft Corporation
> This posting is provided "AS IS" with no warranties, and confers no rights.
> Are you secure? For information about the Strategic Technology Protection Program and to order your FREE Security Tool Kit, please visit http://www.microsoft.com/security.
> Microsoft highly recommends that users with Internet access update their Microsoft software to better protect against viruses and security vulnerabilities. The easiest way to do this is to visit the following websites:
> http://www.microsoft.com/protect
> http://www.microsoft.com/security/guidance/default.mspx
>
Subscribe to:
Comments (Atom)