What is the priority for 1C?

Have a terminal server under win2008 (desktop machine, i7 processor, 4 Gb of RAM — a fairly simple machine for 5-6 users), the users of 1C-Rarus, the database file is located on the same server.
1C awfully stupid, especially when you start and when something is formed.
Moved database to another machine with raid 0 to test the speed.
And gain a single gram, especially measured: the second ended in a settlement (something formed).

That, it turns out that the exchange rate of the screw is not the most important parameter for 1C?

Then what is critical to 1C?
October 8th 19 at 03:28
10 answers
October 8th 19 at 03:30
Why the dB file, and not SQL?
October 8th 19 at 03:32
We should start with the monitoring of resources and not mindlessly carry on different raid, increase OP, etc.
What would you recommend? - waylon.Turcotte commented on October 8th 19 at 03:35
Make in the Configurator measurement of performance for those reports that blunt. Maybe help just code optimization. - Santa_Marvin66 commented on October 8th 19 at 03:38
October 8th 19 at 03:34
Go to the client-server version of 1C.
October 8th 19 at 03:36
Curve code does not help any server performance
October 8th 19 at 03:38
1) Put the server 1C.
2) Rewrite the bottlenecks.
And RAM put 16 gigs. - waylon.Turcotte commented on October 8th 19 at 03:41
October 8th 19 at 03:40
1. Add RAM (8, better 16 GB)
2. Buy SSD 128, split into 2 discs on 1 disc system (not to forget to make the image), the 2nd 1C.
3. Bought a conventional screw, such as WD Green. From the disc 1C being made daily backups.
Such a system should work just fine.
October 8th 19 at 03:42
the simplest open task Manager and view CPU usage and memory. If memory is not enough and everything in the swap then put the memory still.
If percent is not enough to change the percent faster, I feel 1C uses only one core to generate the report.
further it is necessary to monitor the screws using perfmon utility
where did you get your рейд0 will be faster on randomly reading I don't know. take one SSD and drive on it.
October 8th 19 at 03:44
No super powerful server will not save you from bad code, 1S not so slow if it correctly to prepare the configuration is written correctly, find a really good programmer, who will find bottlenecks in your configuration and will propose what and how you can improve. 5GB is not so much the base, to very slow.
October 8th 19 at 03:46
In principle, the file mode faster SQL-tion, but a small percentage of users < 5.
If the DB file and are actively working about 10 — the brakes certainly will be, even on the basis of 1.5 GB.
Ie is clearly on the application server+SQL. You can use MS you free postgres and DB2, but they are slower MS 20-30 percent.

On gland — like Pavlo critical memory and disk. To drive a classy version of the SSD screws of the mirror (just be sure to mirror, and then statistics on the reliability of there almost)
The problem was unoptimized additional notes :( just solved... - waylon.Turcotte commented on October 8th 19 at 03:49
Yes, this happens — sometimes a single index in the values table can solve the problem - Santa_Marvin66 commented on October 8th 19 at 03:52
about two months ago from two different clients, flew SSD on servers, worked mirror. operating time was on average about 1 year for one client and about 1.5 years for another client, one works in the housing sector (SSD worked for about 1 year in the accounting Department), another client in the sphere of the social.services (accommodation for people with disabilities and veterans), also the server for accounting. OK, the backups were not fatal. But the fact nifiga not happy. - nicole.Parker commented on October 8th 19 at 03:55
October 8th 19 at 03:48
For file 1C top priority time random access to the file!
And ready not to reduce it, on the contrary even.
This tunim caches the file system on the server, and brains moving in the direction of the raptors and SSD :-)
Well, to clean up the grid. A situation where the unmanaged switch, the switch sits and switch whipping, unacceptable!

From personal experience, the old core 2 duo with 4 gigs operatives and a black woodpecker was enough to delight the five bukhov, file 1C.
1C and a file on the RAM-disk, all burning with Napalm...

PS. In General, and SQL server, is also possible and need to know how to cook!
Oh, and please tell us more about the results of the operation of system 1C in ramdisk? tests any conducted to compare? how often, where, and how backups are removed? was there a disk are duplicated somehow on-line is physically on another machine in case of a bsod? how much time the system is operated, whether there were failures, as was decided? - waylon.Turcotte commented on October 8th 19 at 03:51
support, also interested in the question of reservation in real time - Santa_Marvin66 commented on October 8th 19 at 03:54
yeah, me too, for a few years does not let a question about DB in memory - nicole.Parker commented on October 8th 19 at 03:57
SADKO - do not take the trouble, share your experience, if "file 1C on the RAM-disk" is the same as the kernel in RAM memory, then at least I know where to dig, with the database are not met. - waylon.Turcotte commented on October 8th 19 at 04:00
I think that Sadko had in mind when the OS creates a virtual disk that entirely resides in memory. It fits the file database.
In the end, are just wild speed access and read files many times faster than a modern SSD, but also a headache: the files on the disk in RAM must be constantly synchronized with some other drive, otherwise BSOD or any other operating system failure (e.g. loss of power) - and data is not there.

Early on ebay flashed such a thing as a Gygabyte i-RAM device, which from memory modules DDR did in fact the same hard drive, but I can't remember it in the OS looked, probably SATA (PCI not taken into account). Now these devices somehow became less... - Delbert_Bauch commented on October 8th 19 at 04:03
Oh-how many questions in total so right, although I have tried the hardware, but the RAM disk was cheap, cheerful and reliable...
Reliable because it is easily copied directly during operation and if something went wrong, return all in the ass...
The fact that the probability of a sudden loss of access to the file is not null, and the file of 1S there a kind of logging records, so you can just copy on live at least every minute.
And the night was done update incremental archive which was copied to the file storage with Raid5. So, in the event of failure of the built in hard, RAM snapshots was written in the vault, and interested persons receive notice. - nikita.Stracke commented on October 8th 19 at 04:06

Find more questions by tags 1CNginxWindowsServers