Performance

We started using Navision 3.60 NL and are experiencing problems (locking and even throwing out jobs) when users are doing there work at the same time?!?! When somebody is making an order entry, we are unable to make a pick or register a receipt or an order. Hardware is not the problem. Seems to me 3.60 is. Anyone with the same problems and maybe solutions?

quote:


Originally posted by harmen lebbink
Hardware is not the problem.


Are you sure? What is the size of the database?, are you using dimensions?, do you have manufacturing? SQL or C/Side? What is your hardware configuration?

Hi David, thanks for your reply; this is the data you need: Our hardware has been tested bij Navision Netherlands according to the standards from Denmark. Database 36Gb (4x9), in use: 25Gb and growing, Cache 900 Mb, Objectcache 16Mb, Commitcache is on. We use no manufactoring, we do use warehouse management (seems to be the biggest problem), native database, Compaq/HP servers, raid, with same backup server, 2Gb RAM, 10 18.2Gb disks per server, servernetwork 1Gb switch; i’m not sure about using dimensions, going to check that.

How is Your disk’s configured and how is the database files set up?

quote:


Originally posted by Lars Westman
How is Your disk’s configured and how is the database files set up?


The database is split over the available disks, i dont know what you mean by database files setup?

Hoi Harmen, We have the same problems you mention (Navision 3.10, native database of 2Gb.), especially when invoicing, running a picking list and running the requisition worksheet.

  • We do batch invoicing with a slightly customized Combine Shipments procedure which takes all posted shipments into account instead of only shipments of customers with the boolean checked.
  • We use a collect picking list which collects all released sales orders with inventory, non-drop shipment and so on
  • We run the requisition worksheet for drop shipments once a day and for inventory repeatedly throughout the day
    In all cases, we ‘suffer’ from locks and in the case of the requisition worksheet even throwing out jobs, as you mention. The bad thing about this: it can’t be solved, unless you are going to use SQL server (then you won’t have table locking, I’m told) [xx(] The less bad news, however, is that your clients’ performance can make a lot of difference. Use fast PC’s for users who execute most batch processes. Or use a terminal server session to the server itself. We use that solution for the requisition worksheet, and it’s much faster, and therefore more reliable and less annoying to other users.

Eeeh, by the way, we do use dimensions, David. Can this be a reason for locking more than average?

What RAID-level are You using (0, 1, 5 etc)?

I checked if we are using dimensions; yes we do, raid level is 1 (not 100% sure, what is the difference?)

The differnce is HUGE. Try searching this forum for RAID and You can se quite a lot of discussions. To make a long story short: You should only use RAID-1 with native Navision DBMS! Put one database file on each pair of disks.

quote:


Originally posted by Lars Westman
To make a long story short: You should only use RAID-1 with native Navision DBMS! Put one database file on each pair of disks.


Hi Lars, We do have a native database with a size of 2 Gb., Raid 1 and 4 harddisks. On each of the two disks I have a database file (DB1.fdb on disk #1 and DB2.fdb on disk #2). Processor is at 1.1 Ghz and we have a memory of 1 Ghz. But we still encounter locking problems on a regular basis…

In your situation, I prefer to RAID-0. Now I will be beat up by Navision sectarians [:p][B)] RAID 0 write perfomance faster than RAID 1 ~ 2 times. Read perfomance faster too. Write perfomance have huge influence to lock time. But why Navison don’t like RAID 0? Answer: RELIABILITY !!! You must care about that by self. But, if you get into a mess with 2 Gb database, it isn’t hardware related problem. Run RAID-0 and wait for hotfix or new version. [xx(] BTW maybe your database is hard customized ?

Sorry David, but Your’e wrong. You will get better write performance if You use separate disks without RAID or with RAID-1. That way Navision will get one commit cahce for each disks. I think we have discused this enough on this forum [xx(]

quote:


Originally posted by Lars Westman
Sorry David, but Your’e wrong.


I Guess you mean ‘Sorry Dalius’, don’t you Lars? [;)]

quote:


Originally posted by Lars Westman
Sorry David, but Your’e wrong. You will get better write performance if You use separate disks without RAID or with RAID-1. That way Navision will get one commit cahce for each disks. I think we have discused this enough on this forum [xx(]


No, one commit cache for file. Only RAID controler knows how many disks on raid. http://www.attotech.com/diamond/pdf/RAIDWhitePaper.pdf

the discussion seems to solve some problems here and there allthough we still have the same problems, in another group i read a message that looked like our problem but with performance in about 1/10 of the time we need. we need to process (shipment/batchpick/invoice/label) about 3000-4000 orders a day (within 13 hours) the average order contains 22 lines. in the same time we receive about 20.000 - 30.000 items a day (about 500 - 1500 lines), we put in around 1000 new orders, we run several jobs for the buying dept and for our financial dept. all at the same time:(

“We started using Navision … 25 Gb and growing” - Cool [:p] Harmen, If seriously, 4 years ago I had Navision implementation with less transactions (~60000 item transactions per day, yours I see ~150000). We decided to use replication due perfomance and remote users. All was on native database. It was only one reasonable solution. Now is SQL server platform. But I don’t think that it will be easy run such system. I think that you need SQL and Navision experts help. Sometimes solution covers more than hardware or software. BTW how many sassions ?

quote:


No, one commit cache for file. Only RAID controler knows how many disks on raid. http://www.attotech.com/diamond/pdf/RAIDWhitePaper.pdf


Sorry. Wrong again [;)]. Navisions native DBMS gives You one commit cache for each physical disk-id. And that is a thing that the OS (and Navision) is aware of. Haven’t You been looking at Your processes slave.exe when You’re running RAID-0? So: If RAID, use RAID-1.

75 concurrent users

Lars, check this one: http://www.mbsonline.org/forum/topic.asp?TOPIC_ID=3946&SearchTerms=slave.exe [:D] Harmen, 75 users - isn’t hardware releated problem. It’s behind Navision native database possibilities.