All geek questions in one place

Conceptual question (e.g. tool) LoadRunner

I am using LoadRunner to stress test a J2EE application.

I have: 1 MySQL MySQL server and 1 JBoss application server. Each of them is a 16-core (1.8 GHz) / 8 GB of RAM.

Connection Pool: The database server uses max_connections = 100c my.cnf. The application server also uses min-pool-sizeand max-pool-size= 100 in mysql-ds.xmland mysql-ro-ds.xml.

I simulate loading 100 virtual users from a “regular” single-core PC. This is RAM-RAM 1.8 GHz / 1 GB.

The application is deployed and used on a local Ethernet network at a speed of 100 Mbps.

I use rendezvous points in sections of my stress testing script to simulate real parallel (rather than simultaneous) use.

Question:

CPU utilization on this bootable PC never reaches 100%, and I guess memory is also available. That way, I could try adding more virtual users on this PC. But before I do this, I would like to learn 1 or 2 basics about concurrency / parallelism and hardware:

  • Using only a single-core load generator, can I really simulate a parallel load of 100 users (each user working with a dedicated PC in real life)? Perhaps the misunderstanding is that 100 threads on a single-core PC will run simultaneously (alternating, that is), but not parallel ... Which means that I can not really simulate the real load of 100 parallel users (per 100 PCs) from single, single core PC! It is right?

  • parallelism: 100- (, , , 100 , ), ethernet concurrency, parallelism Ethernet, . , , ( parallelism) ( 1 ), . , , , , !

  • , parallelism (- "" ), , concurrency, , , () , ? () ? (c) ( ) , , ?

,

/HS

+3
4

Ethernet, , -, HTTP (S), TCP-, , , . IP, IP- Ethernet-, . UDP, Ethernet-, 100 100 1Gbit ethernet-, , .

, . SOAP, 7

+1

, . . ( ) , 100- .

1./2. , . , , - ( HTML, , javascript , ). unicore ; 100- .

- /fancier ethernet- CPU, , .

3. . , 100- 5 Craigslist.

+1

, , , , , "" - 100 ( 100 ), 100 . , , , .

, (.NET/#), asyncrhonous - , , HTTP . 3,0 4 , ~ 2200 . Gb, Gb NIC. , , , > 30% . , - , " ".

50 000 , . 2,8 , 64- 12 . , , 50K.

, , ( CAT 3 CAT 5 CAT 6), , , , , , , .. .. .. " ". , 500, 750, 1000 .

, ... - .

, ~ 7 800 URL . 3,0 (Windows Ultimate 7 64- ) 24 , 28 . Parallel.ForEach() < 5 . , , 20%, 14% (CAT 5 NIC Gb Gb- T-1).

, , IO.

+1

, , , - , , . - , , , , ..... - "" (sp? )

, 100 , . , , 100 , 10 9:00 , .

, , , №1 . 1500 , 15 ( Mercury 1 1996 ).

-SQAForums WinRunner, LoadRunner

-YahooGroups LoadRunner, Advanced-LoadRunner

-GoogleGroups lr-LoadRunner

-Linkedin LoadRunner (), LoadrunnerByTheHour ()

Aluminum Mercury (1996-2000)

CTO, Newcoe Performance Engineering

0
source

Source: https://habr.com/ru/post/1788119/

More articles:


All Articles