Question on SQL Cluster Resource Groups

I have a sql cluster with my virtual server name, IP address, etc. in one
group. In the other group I have my sql instance, san attached disk
resources, sql agent etc. In this second group I also have a ftp server that
utilizes the vbscript for fail over and uses a san disk resource.
The problem I have is if the second group fails over say, due to a sql
server error which has happened, then my ftp site dies as users put to the
ftp site via the name / ip of the virtual server but then the virtual server
cannot see the disk resources that are part of the second group.
Is there any issues with putting the virtual server name & ip address in the
same group with sql, sql agent, disk resources, etc.? In essence, putting
everything in one group.
Thanks in advance,
John
You have a couple of options, and I'm sure different people may have
different opinions w.r.t. their pros and cons.
Option 1:
Move the resources from group 1 into the second group, and edit the
dependencies property of your virtual server to include the disk resources
you mentioned. In order to protect the HA of your SQL Server instance, you
may want to edit the advanced property tab of the resources--moved over from
group 1--so that their faulire doesn't affect the group.
Option 2:
Keep the groups as they are. Create a file share resource(s) in group 2, and
use the file share via the virtual server. I don't know the nature of your
particular virtual server, so I'm not sure this works for you.
Personally, I prefer to keep the SQL Server group clean, and will try very
hard to question any attempt to include additional resources in that group.
However, in your situation you already have a bunch of non-SQL Server native
resources in the SQL group so adding a few more may not do a great deal more
potential harm (of course that all depends on the details of the virtual
server in question.)
Linchi
"John - PDX" <JohnPDX@.discussions.microsoft.com> wrote in message
news:D5396113-A4A6-4EA3-973F-B4CBF2021CDB@.microsoft.com...
>I have a sql cluster with my virtual server name, IP address, etc. in one
> group. In the other group I have my sql instance, san attached disk
> resources, sql agent etc. In this second group I also have a ftp server
> that
> utilizes the vbscript for fail over and uses a san disk resource.
> The problem I have is if the second group fails over say, due to a sql
> server error which has happened, then my ftp site dies as users put to the
> ftp site via the name / ip of the virtual server but then the virtual
> server
> cannot see the disk resources that are part of the second group.
> Is there any issues with putting the virtual server name & ip address in
> the
> same group with sql, sql agent, disk resources, etc.? In essence, putting
> everything in one group.
> --
> Thanks in advance,
> John
|||I think your biggest problem is one of definition. By definition, each
Cluster Resource Group IS A SEPARATE VIRTUAL SERVER.
Each group has its own Disk, IP, Network Name, and application resources.
If the FTP server is dependent on the SQL Server instance, then define the
FTP site resources within the SQL Server Cluster Resource Group; otherwise,
create a separate Cluster Resource Group for IIS.
Every Cluster has at least 1 resource group BEFORE there exists any failover
applications, the Cluster Group itself, which must stay independent and
uncontaminated by other resources.
Cluster Resource Group:
Quorum Disk, Cluster IP, Cluster Network Name.
Microsoft Distributed Transaction Coordinator (MS DTC) Resource Group:
MSDTC dedicated Disk, MSDTC IP, MSDTC Network Name.
SQL Server Resource Group:
SQL Server dedicated Disk, SQL Server IP, SQL Server Network Name, SQL
Server Instance, SQL Agent.
Internet Information Server (IIS) Resource Group:
IIS dedicated Disk, IIS IP, IIS Network Name, FTP Site.
Although possible to include the IIS resources within the SQL Server
Resource Group, you would be better off creating a separate group dedicated
to IIS resources. End users would connect to SQL Server through the SQL
Server Cluster Resource Group Network Name and SQL Server Instance name
defined within that group. End users would connect to the FTP site through
the IIS Cluster Resource Group Network Name and FTP site name defined within
that group.
Whenever a Cluster Resource Group fails over, the Disk, IP, Network Name,
and dependent application resources defined within that group fail over
together. Even if the SQL Server resource group failed over, the FTP site
would still be available on the original host through the IIS Cluster
Resource Group Network Name. They are independent.
You use the physical node names to define cluster resources and to manage
non-clustered components on each node, like the OS and swap spaces, Network
Backup software, etc. You use the Cluster Resource Group Network Name to
manage the cluster as a whole. You use the MS DTC Network Name to manage MS
DTC, the SQL Server Network Name to manage SQL Server, and, finally, the IIS
Network Name to manager IIS. All independently but from within the same
cluster.
To go one step farther, you should really install separate hardware and
cluster IIS using Microsoft Network Load Balancing high availability
clustering, not Sever Cluster failover clustering.
Hope this helps.
Sincerely,
Anthony Thomas

"Linchi Shea" <linchi_shea@.NOSPAMml.om> wrote in message
news:OqN4Qps0FHA.2932@.TK2MSFTNGP10.phx.gbl...
> You have a couple of options, and I'm sure different people may have
> different opinions w.r.t. their pros and cons.
> Option 1:
> Move the resources from group 1 into the second group, and edit the
> dependencies property of your virtual server to include the disk resources
> you mentioned. In order to protect the HA of your SQL Server instance, you
> may want to edit the advanced property tab of the resources--moved over
from
> group 1--so that their faulire doesn't affect the group.
> Option 2:
> Keep the groups as they are. Create a file share resource(s) in group 2,
and
> use the file share via the virtual server. I don't know the nature of your
> particular virtual server, so I'm not sure this works for you.
> Personally, I prefer to keep the SQL Server group clean, and will try very
> hard to question any attempt to include additional resources in that
group.
> However, in your situation you already have a bunch of non-SQL Server
native
> resources in the SQL group so adding a few more may not do a great deal
more[vbcol=seagreen]
> potential harm (of course that all depends on the details of the virtual
> server in question.)
> Linchi
> "John - PDX" <JohnPDX@.discussions.microsoft.com> wrote in message
> news:D5396113-A4A6-4EA3-973F-B4CBF2021CDB@.microsoft.com...
the[vbcol=seagreen]
putting
>

Question on repeated query

I have to select a number of fields for a report. Name Address, city,
state zip, etc. from a list of customers.
I have to select 2000 records from each of 29 states.
Is there a way to do this without repeating and repeating the code in
QA, or by changing the state manually and running it each time?
I have to select 2000 customers from WV, 2000 from VA, 2000 from FL,
2000 from TX and so on.
It's more of just a random selection of customers, with no other
requirements other than the name & address are a good one (i.e. non-null
last name, good zip code, etc).
Any help appreciated.Try,
select
Name Address, city, state, zip
from
t1 as a
where
customerid in (
select top 2000 customerid
from t1 as b
where b.state = a.state
)
go
If you run this query again, do not expect to pull same result because I am
not using "order by" clause.
AMB
"Blasting Cap" wrote:

> I have to select a number of fields for a report. Name Address, city,
> state zip, etc. from a list of customers.
> I have to select 2000 records from each of 29 states.
> Is there a way to do this without repeating and repeating the code in
> QA, or by changing the state manually and running it each time?
> I have to select 2000 customers from WV, 2000 from VA, 2000 from FL,
> 2000 from TX and so on.
> It's more of just a random selection of customers, with no other
> requirements other than the name & address are a good one (i.e. non-null
> last name, good zip code, etc).
> Any help appreciated.
>