Navision Attain 3.10 client, extremely slow....

joliver

First excuses about my poor english.

Hi have one instalation of Navision Attain 3.1 into a Cluster 2*(Compaq DL380 G2 PIII 1,266Mhz, 1 Gb Ram, 2*18gb RAID1 OS HDD) with one Smart Array of 6 HDD 36GB in RAID10. The server have 400000 cache size and are activated de commit cache, the db ocuped size is 3.5Gb and the total size is 14GB, the O/S server are W2000 Advanced Server, and the clients are running into a w98 machines with PIII 1Mhz with 128 MG RAM, and 8000 cache and 8000 DBMS for the navision client.

The problem is this, with the client navision hi need 3 hours for validate comercial issues (conditions?), and hi have unistalled the cluster and installed a TS (joint's no work), and running the navision client continue with 3 hours and with the TS client no more of 1/2 hour, the time difference is many important.

Hi need to use the cluster service, and not is possible work with the TS, how test and tune the client speed.

the cache size is correct?

How configure the database?

Thanks and another excuses about my bad english, never have studied this language.

Savatage

14gb db with only 3.5 used? Why have it expanded out so far?

joliver

Savatage wrote:

14gb db with only 3.5 used? Why have it expanded out so far?

First, thanks for reply...

Yes, is expanded so far, not make changes at installation at the moment.

Opps, the max. number of concurrent clients is 6, and the max. clients installed are 10.

Another excuses about my poor english, sorry.

wonmo

Increase the RAM on the client machine to at least 256 Megs. You will notice much improved performance.

matteo_montanari

wonmo wrote:

Increase the RAM on the client machine to at least 256 Megs. You will notice much improved performance.

It is a good advice, but it must also increase the client object cache.
I think that the bottle neck is the lan connection... this would explain the difference of performances with TS

Try to increase the server DBMS Cache to the maximum free system memory available.

Matteo

joliver

This morning hi have visited the installations and hi have surprised.....

the client machines aren't PIII 450 with 64Mg of RAM.......


Tomorrow make a test with a Celeron 2.6 with 512MG Ram, and expose the results...


Tomorrrow more...

joliver

Well, today hi have tested the system with a Celeron 2.4 with 256Mg RAM and 50MG Caché, and a DBMS of 170000, and the performance are equal, extremely slow. Snifff....

Any sugestion?

wonmo

Try changing the DBMS cache to at least 500 Meg and make sure that commit cache is checked.

joliver

A question, the DBMS caché its configured at the Attain client, this is for the server?, is for the client cache at the server and is it global for all the clients?, is individual for each client and is it at the server?.... many questions ufff...

i need more information....

Thanks

jsrark

Have you tried running a clinet on the server box and checking performance?
If not then try this to see if this a hardware issue or a software issue.

joliver

The client at the server are faster, the client at the TS are faster, the client directly to the server are slowly.

How check problems with the comunications with the server directly from client? because with TS is faster.

At the server hi have installed the old Navision Financials Server 2.6 with one historic DB, is this a problem, how check.

Thanks.

jsrark

First issue is W98 - not recommended for navision Clients.

joliver

jsrark wrote:

First issue is W98 - not recommended for navision Clients.

Ok, but no think this is the problem, today hi have tested the system whit these results:

Operation: Uncharge one TPV

PIII 450Mhz 64Mg RAM W98 (8000 Caché - 8000 DBMS)
-Client Navision - 5''54
-Terminal Server - 1''12

Celeron 2.4Ghz 256Mg RAM WinXP (30000 Caché - 170000 DBMS)
-Client Navision - 5''12 :shock:
-Terminal Server - Untested


I think is a problem of comunications into the client-server rule, and not have idea how test this.

The lan connection max. use is 34% and normal use is 16%. is a 100MB Lan.

The server cache are increased to 700000 of caché.

:idea: Any suggestions? :idea:

Thanks

Savatage

all your NIC's are 100's? Server too? I have heard that 1000/s nic in the server stops bottlenecks. Our server is 1000 so I have not downgraded to test this myself. Maybe someone else can chime in on this possibility?

joliver

Ok, is a test pending, if this is the cause not problem, and ¿if i upgrade the two server with a 1000's NIC and change switch (not say if it is upgradable), and this not solve the problem, i'm sure it is a configuration problem and before change the hardware i need confirm this situation.

This morning hi have speak with a worker and are comented this, 2 years before, before install the actual cluster-server, temporally aren't installed a normal server, and all the process ist more faster's. This is premonitory of that cluster-server is bad configured or have a hardware problem.

Thanks.

craigbarbs

Hi,

With only 6 clients running concurrently, there should not be too many performance issues. I don't know too much about hardware, etc. A 3.5gb is relatively small. In my view, you should not be having any performance issues. Perhaps it is a development issue.

How many transactions are being processed a day, and what type of transactions are these? What type of validation is being done? In the validation that you are doing, are you reading through records? If so, are you using a good key? If in your validation, you need to read through a number of records to calculate certain values, then maybe you should substitute this with the use of SumIndexFields.

If Navision displays "Searching for records...." in the bottom left hand corner of the status bar then you are almost certainly using a bad key.

I hope that this helps.

joliver

PIII 450Mhz 64Mg RAM W98 (8000 Caché - 8000 DBMS)
-Client Navision - 5''54
-Terminal Server - 1''12

Celeron 2.4Ghz 256Mg RAM WinXP (30000 Caché - 170000 DBMS)
-Client Navision - 5''12
-Terminal Server - Untested


I think is a problem of comunications into the client-server rule, and not have idea how test this.

The lan connection max. use is 34% and normal use is 16%. is a 100MB Lan.

The number of record and the good key not is the problem (i think), with the TS 1 minutes 12 seconds.... is correct for my, the problem is this, the same process with the same number of records and the same key is 491% more slowly with the client conection, and the lan not is satured...


The client-lan-server is slowly (the connection is Client-Server)
The TS-lan-client-server if faster (the connection is TS-Client)

And the lan work correctly,
¿how test the correct port configuration?
¿how check the correct comunication Client-Server?


Thanks.

jbolks

Did you find any solution for the problem...?
We have a customer with the same problem.
jbolks@watermark.nl

Tommy_Schou

What kind of LAN are you running? Is it 100% switched. If not maybe one of your clients has a bad NIC and this causes a lot of retransmissions due to corrupt packages on the LAN.

Just a suggestion!

jbolks

No, even on 1 pc with client/server installation the same performance difference between connection via server or directly a client opening the database.

joliver

Well, i have solutioned the problem, i have reinstalled the W2000 Advanced Server with clean installation, and have uninstalled one Financials 2.60, i sense that is the problem, with one Attain and one Financials at the same server, the services and the host information are good and the work not.

Sorry my english is truely bad.