Problem with reference dimension & snowflaked tables

OK, this is a bit complicated, so I hope I can explain it clearly. I'll try not to get bogged down in the details of the problem domain, but bear with me.

We have a cube with a regular dimension and a reference dimension. The reference dimension is connected through the fact table through the regular dimension (there are many other dimensions as well, but I'll simplify the situation here). Both the regular dimension and the reference dimension have hierarchies that that include as their top level a level called "System". For both of the dimensions, the "System" level is snowflaked in from the same table. So the regular dimension table contains a column called System_key, and a corresponding attribute. Similarly, the reference dimension table also contains a column called System_key, and a corresponding attribute. Both keys point in to the same table. Naturally, the regular dimension table also has a column that points to the reference dimension table, as you would expect with a reference dimension.

OK, I hope that's clear (if a little pathological). Essentially, the "System" attribute is present on both dimensions, and both dimensions get that value by snowflaking in to the same table.

So the problem comes when looking at measures broken down by System. The regular dimension has a "System" user hierarchy which contains a single level, the System level. When I drag that hierarchy on to a pivot chart, I get the wrong results.

From what I can tell, when processing the dimension it is populating the system attribute for the regular dimension based on the key value stored in the reference dimension table, not the regular dimension table. I have no idea why it would do that. But I look at the SQL statement generated during processing of the key attribute for the regular dimension, and it includes in the where statement a join between the reference dimension System_key column and the System table, rather than the regular dimension System_key column as I would expect. The result is totally wrong values.

Does this make any kind of sense? Any idea what could be causing this, and what might fix it (note that totally redoing all the incestuous table relationships at this point is probably not an option).What if, in the Data Source View, you create another instance of the "System" attribute table, using a simple Named Query like: "select * from System"? Then you snowflake one instance with the regular dimension table, and the other instance with the reference dimension table. My guess (maybe someone can confirm this?) is that, unless you explicitly use role-playing dimensions (like Date in Adventure Works), a single snowflaked table in the DSV is interpreted as a single instance, even if it's incorporated into the schema of multiple dimensions. And in your case, I'm assuming that breaking the "System" attribute out into its own role-playing dimension is not an option?|||Deekpak,
Thanks for the suggestion. That does seem to solve the problem.

However, it does seem like a bit of a hack. It would be nice to know if this is a bug in SSAS that may be fixed in the future, or if there's something fundamental that I don't understand here. We snowflake in this table in multiple other dimensions as well (though none others with reference dimensions currently), so it would be good to know if we're setting ourselves for more trouble later.

Any 'Softies out there care to comment?

problem with query

Hi

I have the following bit of code

string test0001 = "Select Max(activity_order) from roomactivitylk";

int max;

SqlCommand cmd15 = new SqlCommand();
cmd15.Connection = con;
cmd15.CommandText = test0001;
max = (int)cmd15.ExecuteScalar();


max = max + 1;

what it does is add 1 to the value max which is taken from the database

however it seems to be set to 0 as everytime it brings back 0 even though the next incrment value should be 2

any suggestions?

cheerts

Hi

I am not sure which parts of the code is in a loop. But i guess, the "int max" declaration should be outside of the loop (if it is already not).

Hope this helps.

VJ

|||

My guess is u want to perform autoincrement . If i m right i think u may not get it because when there r no records exist check the return value.u may get null if u get null make it as 1 or increment with the max value. Plz reply to me am i rt or wrong

Thank u

Baba

Please remember to click "Mark as Answer" on this post if it helped you.

|||

Hi

thanks for the replys manmaged to sort it

this is the soltuion

string maxquery = "Select Max(activity_order) from roomactivitylk ";
//cmd14.ExecuteScalar();

int max;

SqlCommand cmd15 = new SqlCommand();
//SqlCommand cmd15 = new SqlCommand(test0001, con);
cmd15.Connection = con;
cmd15.CommandText = maxquery;
max = (int)cmd15.ExecuteNonQuery();


max = max + 2;
++max;
max++;

ps yes i was tring to increment:-)

cheers!!

Problem with peer to peer replication across a VPN

We are experiencing some problems with peer to peer replication configuration across a VPN between a 32 bit and 64 bit SQL server 2005. We have been able to configure it but we notice that on successful completion of the configuration the 64 bit machine does not have a subscription for the 32 bit machine article. But both servers have articles created and only the 32 bit machine has a subscription. This is a bizarre experience because the configuration was successful.

We know the server windows account has administrative rights at both end and we can telnet each server successful and all the necessary ports are open.

Might feeling is that the firewall or VPN us acting up but I cannot prove it..

I need help or advice about how to resolve this…

I assume you configure the P2P using P2P wizard. When you go through the wizard, in the summary page, do you see replication indiciate it will create subscription from 64bit machine to 32 machine?

Peng