Anyone got a faster way to remove and add a record with partial reload?

[amien]

I used the standard "Test Script" in QlikView to load 5.000 records.

Now i want to remove 1 TransID (with all the lines) and add a new version of the TransID. Using Partial Reload

// Get list with records that needs to be deleted
t2:
ADD
ONLY
LOAD * INLINE [
_TransID, Delete
5000, Y
];

// Get list of all existing TransID Excluding TransID's in t2
t3:
ADD
ONLY
LOAD Distinct
TransID
RESIDENT Transactions
Where Not Exists(_TransID,TransID);

// Remove TransID's which are in t2
Right Keep (Transactions)
t4:
ADD ONLY
LOAD
*
Resident t3;

// Dropping PartialReload Tables
If IsPartialReload() = -1 Then

Trace Partial;
Drop Table t2;
Drop Table t3;
Drop Table t4;

End If;

// Concatenate new version of TransID
Concatenate (Transactions)
Add ONLY
LOAD * INLINE [
TransID ,TransLineID, Dim1 ,Dim2, Dim3, Expression1, Expression2, Expression3, Num
5000 ,1 ,Z ,z ,N ,777, 0, 0.666, 6
];

See also attachment.

My question : anyone got a faster way to do this? Perhaps there is a faster way to great the t3 table or remove the record from the "Transactions" table.

Thanks in advanced

[mariusz_kumansk]

Have you ever used QVD files? Maybe the best solution for you will download your transaction table in QVD file and then you can fast manipulate this data??

[amien]

That's what we do now. it's too slow. it's mil of records.

[mariusz_kumansk]

Did you try to replace the right keep with a exist?

[amien]

that where exists needs to be on my main transaction table. This is slow. It contains alot of columns.

or perhaps i don't quite understand.

How would you please this script with a 'exist'?

Right Keep (Transactions)
t4:
ADD ONLY
LOAD
*
Resident t3;

[marcus_sommer]

I'm not sure if a partial loading is the best approach for a fast loading of data whereby I have no real own experience on the matter - only playing a bit with it some years ago but never used it in real scenarios. If I interpret the various answers to load-performance here in the community right it's really rarely used which makes me think that's like a preceeding load another and more or less convenient way of loading data but not the fastest one.

Beside this I think that your approach contained a detour by reading all TransID's and using them within a keep-statement as filter. I would probably use the following (without partial a load):

temp: load 5000 as TransID autogenerate 1;

table: load * from/resident Source where not exists(TransID);

concatenate(table) load 5000 as TransID, ... other fields autogenerate 1;

drop tables temp;

The important point is to keep all the loadings optimized which is only possible with a where-statement with exists() and a single parameter on it. As far as there are two parameters or other transformations a load needs to be processed record by record and will be quite slow compared to the optimized way.

- Marcus