Showing posts with label analysis. Show all posts
Showing posts with label analysis. Show all posts

Friday, March 30, 2012

Load Balancing SSAS Servers

Hello,

We are trying to load balance a pair of SQL Server 2005 Analysis Services query servers using hardware load balancing (Foundry switches) but we don't seem to be able to get it to work.

The main problem as far as I can tell is that we can't get the SSAS instances to listen on the IP address of the loopback adapter which is configured on both servers.

The netstat -ano command shows that the SSAS instances are only listening on the IP address of the physical NIC in each machine.

We are using a custom port and I can see where this custom port is defined in the msmdsrv.ini file but I can't see anywhere where you can tell SSAS the IP addresses on which to listen.

Should SSAS listen on all available IP addresses when it starts up ? We have re-booted both boxes but this hasn't made any difference ?

Can anyone shed some light on this please ?

Regards,

Paul Lynch

Hi Paul, I have not done anything like this before, so this is a bit of a long shot. I was wondering if the security setting that you need to change in order to use application names might need to be changed. see http://support.microsoft.com/kb/888526 - this article is for AS2000, but I think the same approach still applies for AS2k5.

I think by default it checks that the name/ip address match each other, so this may help.

|||

Hi Darren,

Thanks for your reply.

We actually found the problem - the network guy had forgotten to configure something called Direct Server Return (DSR) on the Foundry so the traffic wasn't being forwarded properly to the SSAS boxes. When I looked into it again I realised that the SSAS process was actually listening on the loopback adapter address as well as the IP address of the physical NIC so the problem wasn't what I originally thought it was.

We are now able to connect to the VIP address on the port specified in SSAS and it appears to connect OK (I can connect from the front-end web server using SQL Management Studio so I'm assuming its working)

However, the KB article you posted has got me thinking that maybe we need to configure an application name for it to work - I'll need to speak to the DBA's because I'm out of my depth with this already - I'm just an IIS guy ;-)

Thanks again.

Paul

Monday, March 19, 2012

List Vs Matrix Vs Table performance.

Has anyone performed any analysis comparing the performance of a List vs. a
matrix and a table (inparticular rendering performance).
Other than obvious things like complexity, and groupings (single groups, row
groups and column groups), has anyone documented the reasons to use each
control.
Thanks
JasonA matrix is for pivot tables/cross tab queries. The list control, at least
in part, is used for form type reports with one record per page or for
free-form placement of controls, and a table is used for columns of data
with repeating rows, which is the most common type of report. Obviously
these are basic comparisions and not necessarily set in stone. For instance
you could use a list instead of a table and get the same results but since a
table already has established columns, rows, footers, and headers with I
don't know why you would.
"Jason" <jasonnotebook@.hotmail.com> wrote in message
news:%23KE2Jk7bEHA.3716@.TK2MSFTNGP11.phx.gbl...
> Has anyone performed any analysis comparing the performance of a List vs.
a
> matrix and a table (inparticular rendering performance).
> Other than obvious things like complexity, and groupings (single groups,
row
> groups and column groups), has anyone documented the reasons to use each
> control.
> Thanks
> Jason
>|||Thanks for the response, but what I am looking for is performance
information.
I understand the usages, just would like to know, given you can do of most
what can be done with one control, using the another, if performance over
development time was my priority, what would I pick.
Jason
"Stephen Witter" <switter@.enpathmed.com> wrote in message
news:%23eaT$8%23bEHA.808@.tk2msftngp13.phx.gbl...
> A matrix is for pivot tables/cross tab queries. The list control, at
least
> in part, is used for form type reports with one record per page or for
> free-form placement of controls, and a table is used for columns of data
> with repeating rows, which is the most common type of report. Obviously
> these are basic comparisions and not necessarily set in stone. For
instance
> you could use a list instead of a table and get the same results but since
a
> table already has established columns, rows, footers, and headers with I
> don't know why you would.
> "Jason" <jasonnotebook@.hotmail.com> wrote in message
> news:%23KE2Jk7bEHA.3716@.TK2MSFTNGP11.phx.gbl...
> > Has anyone performed any analysis comparing the performance of a List
vs.
> a
> > matrix and a table (inparticular rendering performance).
> >
> > Other than obvious things like complexity, and groupings (single groups,
> row
> > groups and column groups), has anyone documented the reasons to use each
> > control.
> >
> > Thanks
> >
> > Jason
> >
> >
>|||I was responding to the second part of your question but neglected to state
that. I personally have never done a comparision between the report types.
I also think that all three are for different uses, so comparing them side
by side seems kind of pointless. Having said that, maybe someone else has a
differnet opinion.
"Jason" <jasonnotebook@.hotmail.com> wrote in message
news:%23YEyXrDcEHA.3988@.tk2msftngp13.phx.gbl...
> Thanks for the response, but what I am looking for is performance
> information.
> I understand the usages, just would like to know, given you can do of most
> what can be done with one control, using the another, if performance over
> development time was my priority, what would I pick.
> Jason
>
> "Stephen Witter" <switter@.enpathmed.com> wrote in message
> news:%23eaT$8%23bEHA.808@.tk2msftngp13.phx.gbl...
> > A matrix is for pivot tables/cross tab queries. The list control, at
> least
> > in part, is used for form type reports with one record per page or for
> > free-form placement of controls, and a table is used for columns of data
> > with repeating rows, which is the most common type of report. Obviously
> > these are basic comparisions and not necessarily set in stone. For
> instance
> > you could use a list instead of a table and get the same results but
since
> a
> > table already has established columns, rows, footers, and headers with I
> > don't know why you would.
> >
> > "Jason" <jasonnotebook@.hotmail.com> wrote in message
> > news:%23KE2Jk7bEHA.3716@.TK2MSFTNGP11.phx.gbl...
> > > Has anyone performed any analysis comparing the performance of a List
> vs.
> > a
> > > matrix and a table (inparticular rendering performance).
> > >
> > > Other than obvious things like complexity, and groupings (single
groups,
> > row
> > > groups and column groups), has anyone documented the reasons to use
each
> > > control.
> > >
> > > Thanks
> > >
> > > Jason
> > >
> > >
> >
> >
>

Wednesday, March 7, 2012

list of databases in Analysis Services

Hi all,

I have to write a program to read the names of databases in Analysis
Services. I don't know which table I can get this information from.

Thanks a bunchHippi wrote:
> Hi all,
> I have to write a program to read the names of databases in Analysis
> Services. I don't know which table I can get this information from.
> Thanks a bunch

would this work:

select dbid, name
from master..sysdatabases
where has_dbaccess(name) = 1

--
David Rowland
http://dbmonitor.tripod.com|||No, it doesn't work. It shows all the databases I have in SQL Server