Hello,

I've been told that PTS performance over a network suffers due to the amount
of traffic generated by PTS, and that Yukon is supposed to solve this
problem. I realize that this is probably an over-generalized question, but
is this true? Or are performance issues that arise from a client accessing a
remote cube more likely to be a result of a less than efficient cube design?

(as an fyi, the app is using ADO MD to access the cube)

Thanks in advance.

rg

I can offer my opinion on this question: It depends a lot ! Since PTS has
client-side cache, the amount of traffic generated by PTS usually decreases
with the lifetime of the session. I.e. at the beginning metadata is
downloaded, then data as needed, but more queries are done - higher chances
that the results for them will be in PTS cache. So I would say that in
typical scenarios average traffic from PTS probably will be smaller then
average traffic for Yukon. However, the biggest change IMO is not in the
throughput, but in latency. Single MDX query could trigger many network
roundtrips in PTS (or it could be zero roundtrips if cache is used), whereas
in Yukon there is always one roundtrip per query. So if the network has high
latency (sattelite link), with Yukon you will get much better
predictability.
Additionally, you can also predict the size of the network traffic in Yukon,
since the result will contain only results for the query, whereas with AS2K,
network traffic may include intermideatery results needed in order to
compute the final result.

--
==================================================
Mosha Pasumansky - http://www.mosha.com/msolap
Analysis Services blog at http://www.sqljunkies.com/WebLog/mosha
Development Lead in the Analysis Server team
All you need is love (John Lennon)
Disclaimer : This posting is provided "AS IS" with no warranties, and
confers no rights.
==================================================

Thank you for your response. I don't know if this will shed some light on
the issue, but here goes.

The part where the cellset is created and opened executes quickly. Here is
a portion of the code (this is VB6):

Set objCellset = New ADOMD.Cellset
objCellset.Source = strMDX
Set objCellset.ActiveConnection = mobjDBCatalog.ActiveConnection

Where it really bogs down is in the portion of the code that retrieves the
data. The following is in a loop which loops over the X and Y positions of
the cellset. The following is not the exact code, but the bottleneck is
referencing the data in the cellset as shown below:

....
var(lngX - 1, lngY - 1) = objCellset(lngX - 1, lngY - 1)
....

For a typical query what happens is, the first 100 or so data points go
quickly. It starts to crawl when I get past a certain point (exact point
varies each time).

Thank you again.

rg


"Mosha Pasumansky [MS]" wrote: