How can I trace the bottleneck of the file 1Сv8.1 over the network?

Question

Computer networks

There is a main office, remote branches and a distributed 1C v8.1 with a volume of> 3GB. The exchange takes place via FTP for lack of other means.
Remote branches have two machines each: PC1, PC2. The machines are:
OS: Windows XP with SRP enabled (i.e. no antiviruses).
Network: PC1 and PC2 are connected via a 100mbps switch and receive DHCP addresses from the ADSL modem.
Network adapters: integrated realtek.
Processors: Celeron ~2ghz, Atom ~2x1.66ghz.
RAM: ~1gB.
Railway: common barracuda 7200rpm
The logical structure of the sections of the railway:2GB - swap file, 15GB - partition for the database, 40GB - system partition, 25GB - partition for backups. The swap partition and database have a cluster size of 64kB.
Access to the database: For PC1, the path to the database looks like: x:\base\ , for PC2 it looks like: \\PC1\base\
All of the above was done on a whim and gave a performance increase on PC1 > 200% relative to the original position.
The main delays occur when searching the database for a product barcode read by a scanner, or a discount card (30 -> 7 seconds after optimization)
At the moment, the situation is as follows: PC is flying, no one complains about five-second delays, but on PC2, delays reach 30-40 seconds when creating a new receipt, when searching for a discount card. The most annoying thing is that at the moment of freezing, I do not see any load on the network and processors. I would like to know practical advice and links to useful information.
I read a lot on the Internet, but it is mainly about optimization for SQL and RDP, which is completely inappropriate for me. They say you can clean the logs every day, I'm also thinking about connecting a network drive so that PC2 looks at the x:\base path - does it make sense?

Answer 1

SQL? expensive? the first link in Google using the keywords "1c postgres" leads to v8.1c.ru/overview/postgresql.htm
postgres, like mysql, is free.
Yes, it will take money and time to transfer 1s to postgres, but it will remove all crutches in the form of FTP, crooked balls, and brakes
and add flexibility to the solution and centralization of storage and backup copies of databases.

Answer 2

Repair the network. On two users (besides such), the 1C file base does not blunt, SQL will not save.

Answer 3

in the debug mode on the “brake” machine, take a performance measurement. And then see what action takes a lot of time.
And only then it will be possible to understand what is easier to do - optimize the code or deal with iron

Answer 4

If you close 1C on PC1, does PC2 slow down?
Look with perfmon on PC1 disk queue

Answer 5

Swap cars. Try to access by ip in the path to the database. Disable shadow copying. Check DNS. Poshaman somehow ...
If you solve the problem - unsubscribe, please.

Answer 6

Tupit, because file 1C over the network. Make the most budget version of the terminal server - unlock the connection via the terminal of several users on win xp with a patch. Connect to PC1 with PC2 Remote Desktop and it will be just as fast.

Answer 7

I am not a guru in 1C, but it seems to me that the bottleneck is RAM: ~ 1gB + access to the database over the network and 100 Mbps each.
I wonder if there is a raid on PC1. + 6 feeling tells me that when accessing UNC PC2 first resolves the network name, gets the listing of shares, loads the database files corresponding to the search area into memory, and only then searches for it.
As an option, you can try to hook \\PC1\base\ with a disk to PC2 and see what happens ... sometimes it saves.
An even more correct option is to get hold of high-performance network storage on SAS and give both machines access to the database via a 4-6 Gb bus.

Answer 8

The most correct option is to switch to SQL, I doubt that there is a plug in the network for the experiment, look at the statistics on the switch or through some programs in local places. And it is necessary to add more memory definitely!

Answer 9

Why don't you want to use terminal mode if SQL doesn't suit you? In general, you will always have brakes for the simple reason that the loading of channels for any provider can be different. At the same time, you will carry more and more data over the network.

Answer 10

And even so, put on one of the Windows 2003 PCs (if you don’t want to open the terminal in WinXR)
, the network is the bottleneck.
File work involves the constant dragging of a database file (or part of it) over the network.
3 gigabytes is not much, but it all depends on what you have filled with these 3 gigabytes.
Get a ToolCD and see what takes up the most space in your database.
But these are all trifles. I repeat the main decision - the terminal!