First, my development machine:
It’s SQL Server Express x64 with Reporting Services running on Win7 with 8Gb RAM.

Then my production machine:
SQL Server with Reporting Services running on Windows Server 2008 with 8Gb RAM (x86 though, so only 4Gb usable).

I have a table that contains about 3000+ records containing polygon coordinates.  The database that holds the table is identical on both machines.  A sample query I’m running on the dev machine executes in a fraction of a second but takes 11 seconds on the production machine.  When I looked at the execution plans, I was surprised to see that they are completely different on the two machines.

The query is:

SELECT *
FROM USCounties
WHERE geography::STGeomFromText('POINT(-86.7535 33.6499)', 4326).STIntersects(Polygon) = 1

On the production machine the execution plan shows 3 steps:
SELECT Cost: 0%
Filter Cost: 97%
Clustered Index Scan Cost: 3%

But on the dev machine the execution plan is much more complex and includes things like “Nested Loops”, “Hash Match” and “Compute Scalar”.  The indexes on the two machines look identical from what I can see and the data is identical.  I would expect a performance difference just based on x86 vs x64, but I don’t understand the difference in execution plans…

I really don’t know much of anything about db optimization, any tips from the experts here?

 - Dave

the ltcResults table have more than 30 million records.

CREATE VIEW [dbo].[vwRptHeaderToResultsFormRpts]
AS
SELECT TOP 100 PERCENT dbo.vwResults.ParentID, dbo.vwResults.BatchID, dbo.vwResults.PersonID, dbo.vwResults.FormName, dbo.vwResults.FieldName,
dbo.vwResults.ValueString, dbo.vwResults.ValueNumber, dbo.vwResults.ValueDate, dbo.vwResults.ValueBit, dbo.vwResults.Active,
dbo.vwRptFields.ControlGroupType, dbo.vwRptFields.ControlCaption, dbo.vwRptFields.ReportCaption AS ReportCaptionField,
dbo.vwRptFields.ControlName, dbo.vwRptFields.ControlSeqNo, dbo.vwRptFields.itemtype, dbo.vwResults.FirstName, dbo.vwResults.LastName,
dbo.vwResults.PersonFriendlyID, dbo.vwResults.DateCreated, dbo.vwResults.IsChildRecord, dbo.vwRptFields.FormCaption,
dbo.vwRptFields.FormSeqNo, dbo.vwRptFields.TabCaption, dbo.vwRptFields.TabSeqNo, dbo.vwRptFields.PanelCaption, dbo.vwRptFields.PanelSeqNo,
dbo.vwResults.ResultsPK, dbo.vwResults.UserID, dbo.vwRptHeader.CrystalRptID, dbo.vwRptHeader.ReportCaption
FROM dbo.vwRptFields INNER JOIN
dbo.vwRptHeader ON dbo.vwRptFields.FormName = dbo.vwRptHeader.RptFormName INNER JOIN
dbo.vwResults ON dbo.vwRptFields.fieldname = dbo.vwResults.FieldName
ORDER BY dbo.vwRptFields.FormSeqNo, dbo.vwRptFields.TabSeqNo, dbo.vwRptFields.PanelSeqNo, dbo.vwRptFields.ControlSeqNo
GO

other views are not taking more time but vwResults

At the same time, this company is packing inventory in boxes in order to fulfill orders. I have a table setup for this in which inserts are made for an inventory id and a box id (tied to a purchase order table) to show when an item has been packed and is no longer in inventory. Roughly 6000 items are being added per day into inventory as well as being shipped out of inventory.

Before an item is being shipped – I have to first check the inventory table to make sure the item exists in inventory and also need to check the shipping table to make sure the item hasn’t been shipped. I’m also checking these tables on the sales side to verify when an order has been completed. These two tables are being hit every few seconds both with reads and writes.

I’m running into some performance issues as the size of my table is growing. I’ve been researching the problem and have a couple of ideas on how to fix this – but wanted to check here first to see if I’m on the right track.

1. Indexing this table in order to get better performance during the lookups. From what I’ve read, this will improve the select queries – but will slow down the inserts since the table will be re-indexed after each insert. Since I’m inserting into these tables every couple of seconds – I’m wondering if having indexes on these tables might degrade performance instead of improving it.

2. The queries are being performed via my application. Ie, a query checks to make sure the item is active in inventory – results are returned to the application and either an error is thrown or an insert statement is re-submitted to the SQL server. By moving this into a stored procedure – I’ll remove the network traffic associated with going back and forth between the application server and the sql server.

3. I’ve read a little about partioning and it sounds like this might help (ie, partioning on each of the location id’s) – but I have not gone into it enough to really know enough about it.

Hopefully my question was clear. Any advice or suggestions are appreciated.

I have migrated a database from 2000 to 2005, updated the statistic and rebuild the index.

When I run the following query
“select top 500 * from myview order by name”, it is extremely slower in 2005 than 2000. The myview simply does an inner join with 2 table. The name is varchar(50). Strangely when I remove the top 500, the time to run the query on 2000 and 2005 quite similar.

Does anyone have any idea why this is occurring, or any input, this is really puzzling me???

Thanks