Quantcast

Large blocksizes

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Large blocksizes

Jori Gielis

Dear Sir or Madam,


 

Me and my colleague are ICT students at Thomas More Geel in Belgium.
This year we are doing a 3 months internship at the company Valcosoft.
The company has given us a subject, this subject is SAN.
Currently we are in our fifth week and we are doing some tests with IO-meter.
We got some question about it and it is hard to find some answers. 
When you need to test a SAN solution with IOmeter you always need to take different block sizes like 4KB-32KB-...1024KB and 4096KB. Is there a particular reason behind it? 4KB is the blocksize of NTFS, but why do we want to test the bigger blocksizes? 
The second question we have is why do you need to test with different values like 100% Read & 0% Write, 50% Read & 50% Write,.... And why can't you use a value like 60%Read & 40% Write? Does 50% read en 50% write happen in real life? 


We hope you can help us.


Thanks in advance.


 

Yours faithfully


 

Jonas Deckers and Jori Gielis


------------------------------------------------------------------------------
"Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
Instantly run your Selenium tests across 300+ browser/OS combos.  Get
unparalleled scalability from the best Selenium testing platform available.
Simple to use. Nothing to install. Get started now for free."
http://p.sf.net/sfu/SauceLabs
_______________________________________________
Iometer-user mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/iometer-user
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Large blocksizes

Rizul khanna

Hello Jori,

Greetings.

I would try to help you with my experience with IOmeter so far.

IOmeter is Benchmarking tool that will give you a rough estimate of how and what to expect from your SAN/ Storage solution at times of stress/ IO demand. The I/O request that your application generates goes to your OS and then via OS through the File System Reaches the H/W layers of the Storage which serves your request. The different I/O options that you see in the IOmeter actually represent the kind of I/O requirement patterns that application normally generate. 

For example, when you open any computer application, it most likely reads blocks from your storage in a sequential pattern, the same as they're normally written on the disk. To understand this better, say you open a MS Word application, the application would normally take less than 5 secs on windows 7 based home-desktop to open, reading several hundreds of blocks of information from the underlining disk. This superb performance is possible coz. the Read of I/O blocks is mostly sequential and not random, whereas if the same no. of blocks were to be read randomly it may have taken a hell lot of time.

It is a highly recommended practice to study your applications/ workload first, understand and note down the size of I/O request it is generating and then size your file system and storage OS block size accordingly for best performance of the underlining storage. When you read on the backside of a regular SATA disk that the IOPs are around 75, please note that these 75 IOPs would only be served by your disk in some specific conditions only, the manufacturer of the SATA disk will never disclose these conditions, the nature of the I/O requests(% of Read and Write), the amount of sequential and randomness of the request under which you'll be able to generate these. There is a high possibility that the same amount of IOPs will never be achieved if your application's I/O request patterns are not aligned with the I/O request parameters(Undisclosed by the manufacturers). 

I hope I have answered some of your queries. Please feel free to revert. You can connect with me on linkedin.

@IOmeter User Subscribers- Please correct me if I'm wrong somewhere.


Warm Regards,

Rizul Khanna



On Sat, May 3, 2014 at 2:38 PM, Jori Gielis <[hidden email]> wrote:

Dear Sir or Madam,


 

Me and my colleague are ICT students at Thomas More Geel in Belgium.
This year we are doing a 3 months internship at the company Valcosoft.
The company has given us a subject, this subject is SAN.
Currently we are in our fifth week and we are doing some tests with IO-meter.
We got some question about it and it is hard to find some answers. 
When you need to test a SAN solution with IOmeter you always need to take different block sizes like 4KB-32KB-...1024KB and 4096KB. Is there a particular reason behind it? 4KB is the blocksize of NTFS, but why do we want to test the bigger blocksizes? 
The second question we have is why do you need to test with different values like 100% Read & 0% Write, 50% Read & 50% Write,.... And why can't you use a value like 60%Read & 40% Write? Does 50% read en 50% write happen in real life? 


We hope you can help us.


Thanks in advance.


 

Yours faithfully


 

Jonas Deckers and Jori Gielis


------------------------------------------------------------------------------
"Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
Instantly run your Selenium tests across 300+ browser/OS combos.  Get
unparalleled scalability from the best Selenium testing platform available.
Simple to use. Nothing to install. Get started now for free."
http://p.sf.net/sfu/SauceLabs
_______________________________________________
Iometer-user mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/iometer-user



------------------------------------------------------------------------------
"Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
Instantly run your Selenium tests across 300+ browser/OS combos.  Get
unparalleled scalability from the best Selenium testing platform available.
Simple to use. Nothing to install. Get started now for free."
http://p.sf.net/sfu/SauceLabs
_______________________________________________
Iometer-user mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/iometer-user
Loading...