Configuring Data Build in Windows Server Performance Monitor

Every experienced system administrator knows that the best indicator of 1C performance deterioration is the chief accountant moving towards the IT department at a speed exceeding 1.1 m / s. But only the wisest of them set up the collection of counters so that this meeting does not catch them by surprise. Let's talk about this under the cut ...







Epigraph:

There are two reasons why a computer can slow down:

1. Virus.

2. Antivirus.

© advice of experienced system administrators



I will not be mistaken if I say that every office administrator faced the question: Why does 1C slow down?



And again, I won't be mistaken, if the first thing he (s) does at the same time, it will open the task manager.

More advanced ones, of course, will configure the collection of Performance Monitor counters (Zabbix in this context is about the same). 



Moreover, there are more than enough instructions, checklists for setting up. This is what scares.

I'll try to offer you an overview of the main ones and my own compilation.



Attention!



The name of the counters differs not only depending on the language of the operating system, but also on its edition.



Let's add to this the vision and mistakes of the authors of the publications and understand that a simple copy-paste may not work.



In the case of perfmon, this will be aggravated by the fact that you will not receive any errors when creating counters in the command line, they simply will not be collected.



In order to see a list of all performance counters available on the current computer, you need to run

• typeperf -q [object] will list all counters
• typeperf -qx [object] will display a list of all counters by hardware instance, for example, separately for drives A: and C: 

Where the optional parameter [object] is a filter on the type of meter, such as PhysicalDisk

This conclusion can redirect to a file and then have to choose from it the necessary

typeperf -qx -y -o counters.txt



In the future, to get summary statistics should be replaced in the event of a key - qx the name of a particular instance on (_Total) , and to get statistics for each instance separately on  (*)



For example:



\ PhysicalDisk (_Total) \ Current Disk Queue Length 

\ PhysicalDisk (*) \ Current Disk Queue Length 



My recommended way is to create a bat file of 3 lines.



logman create counter 1C_counter -f bincirc 

logman update counter 1C_counter -cf assembled.txt 

logman update counter 1C_counter -si 15 -v ​​mmddhhmm 



And add the names of the counters to the assembled.txt file. One per line. A working and recommended example for Windows Server 2012 R2 ENG will be below.





Actually, hurry-ups may not read any further. They don't even read anymore.

With the rest we will deal with the recommendations of the best dog breeders



Let's start by studying the advice of the vendor himself: microsoft.com

Publishing Windows VM health





Using this option, you definitely won't be mistaken, but it contains counters that are not quite necessary for monitoring the 1C server.



Further, but rather higher, in my top of options there is a recommendation from Evgeny Valerievich Filippov

1C Handbook: An Expert on Technological Issues. Edition 2 The



list is small, but everything is on the case and it is clear that the author used it in his work.

 





The list of the book Methodological manual for the operation of large information systems on the platform "1C: Enterprise 8"

A. Asatryan, A. Golikov, A. Morozov, D. Solomatin, Y. Fedorov is

even more concise, it added monitoring 1cv8, ragent, rphost, rmngr I will put it on a separate list, because it may and probably will not interfere with any option, except for spaced SQL and 1C servers.





List of equipment counters.

 



Next is an article with ITS Analysis of hardware utilization for Windows Elena Skvortsova and its full copy on kb who has access there , it describes the entire setup process in detail and with pictures. This is very useful for the first setup.

For all the usefulness and accessibility of the article, the feeling does not leave that it was written like the famous letter from Matroskin: "your son is uncle Sharik", different people. For example, the text does not match the pictures, for some counters, threshold values ​​are described, but they are not in the list, some counters in the list are doubled, because of this, it will not be possible to copy-paste logman in the command line. This is just a little discouraging for beginners.







Lyrical digression: Not even a month has passed since the regional round of the ITS competition, where one of the questions was composed in this way, there is only one answer in the code, and in the picture, the mathematically correct one is completely different. The organizers relied precisely on the correctness of the code. Although it is understandable that the code is poorly checked, all scientific books warn about this in advance.

 



The list is closed by foreign vendor agents . 

www.veritas.com  Analyzing SQL Performance using Performance Monitor Counters



It is clear that they have never heard of 1C, but the fact that they saw an order of magnitude more servers is a fact.

 



red-gate.com

SQL Server performance and activity monitoring



 



As for counters for MS SQL, my list was at the beginning of the publication.



There are an incredible variety of options, as well as experts (it's not a fact that now one of them hasn't moved quietly under the table at the sight of him).

However, a real cheeky admin will never show his attitude, he will play with his eyebrows at the most and will listen to the server's music.



Those interested can spend a couple of winter (summer) evenings sorting out the full list.





- Navigator, instruments!

- Fourteen.

- What is fourteen?

- And what, devices !?

© www.anekdot.ru A



vigilant reader will say: It is not enough to collect equipment counters, you also need to analyze them.

And I'll show him this table.



Technical digression: Although it expresses the opinion of respected experts, it must be treated with understanding.



For example, many will remember the times when they begged the director to buy additional 32 MB for the server of the above-mentioned accounting department. The same goes for the speed of the discs. These values ​​are deprecated.



Attention!



" ". . . , .





Perhaps you have your opinion about equipment monitoring, come to the comments, write your thoughts, preferably with links to sources of knowledge.