>>>Hi,
>>>Thank your for your reply
>>>
>>>1. Network : 10BaseT AND Ethernet
>>>2. my app User : 5
>>>3. my app is order entry system ( customer Data is about 13,000 rec and order data is about 15,000 rec )
>>>4. when I using index(CDX) in my DBC , my app is same speed -> Slow :-(
>>>5. my query -> "=SQLEXEC(nCon, "select cstno, cstnm, conm, copart, cotitle from customer where cstnm like 'choi%'", "tmp")
>>>
>>>Pls help me
>>>
>>>ASAP
>>
>>1. SQL doesn't optimize when you use a wildcard character like the "%".
>>
>>2. Do you have an index on the cstnm field?
>>
>>3. Could you re-write the SQL code so the WHERE clause is ....WHERE cstnm = "choi" OR cstnm = "tmp"
>>This would allow an index on cstnm to optimize the code.
>>
>>HTH
>>Barbara
>
>Barbara:
>
>You mis-read the SQLExec parameters. "tmp" is the 3rd parameter which is the cursor name to which the results are to be returned.
>
>Anyway, with that few records, I would not expect performance to be slow at all. We use true networks, and I have never been impressed with peer-to-peer networks such as what he is doing. I think that is the main reason for problem, especially if the "server" is also used as a workstation (which it probably is).
>
>Our general policy is we will never support or set up a peer-to-peer network for this very reason because users/customers rarely ever blame the hardware/OS/Network, they always blame the application and the programmer.

Hi Mark,
Wouldn't agree with peer-to-peer networks are slow. I also use a peer-to-peer network on some sites and if machine population is low it's more convenient and cost-effective. Why would I keep a dedicated server for only 5-6 machines ? Why not blame here the hardware or even SQL pass through setting ? My applications run on peer-to-peer without any performance loss compared to dedicated systems in small environments. Here a dedicated server would do nothing more than sharing a harddisk in fact and it wouldn't use even 30% of CPU time for 5-6 machines. Why pay for that extra 70% ? Any peer-to-peer machine could handle it too. Instead of SQL pass through directly using the data from a mapped drive could prove effective. On a peer-to-peer network with NETBEUI data transfer rate is about 1.5 Mb/s which is faster than many old local harddisk transfer rates. And besides as you said performance loss with that few records is not an expected thing. I wonder if the code runs fast on local. If it is than network performance tests could be done to see if 10BaseT is working correctly (ie: grounding, misswiring etc). It's so possible that Tx or Rx pairs be under EIA/TIA 568 standarts. I don't expect a performance increase in this case by using non peer-to-peer net.