hello good day,
I just tried primo cache lately and enable the defer-write, but i got a problem when i try to monitor the disk activity using per-mon, I am still seeing a constant drive writes even thought i set the defer-write to 1800sec and still have enough free cache, i also try native and intelligent write mode.
please look the attached files.
any suggestions about this, thank you.
defer-write
defer-write
- Attachments
-
- Untitled.png (113.45 KiB) Viewed 6372 times
-
- permon.gif (32.83 KiB) Viewed 6372 times
Re: defer-write
It seems that there's no problem in your case. The yellow line "Avg. Disk sec/Write" doesn't indicate that the disk was writing. The blue line "% Disk Write Time" and the green line "Disk Writes/sec" exactly shows that writing occurred at intervals.
Re: defer-write
Thank you very much for pointing me out, but I have another concern. I dont know if it is a bug, but i think caching is not working in windows libraries.
Here is the scenario:
I have two libraries, we will named it File and Backup, they are pointed to a different folder on the disk where read/write cache is enabled.
now from the File library i copy a folder containing a mixed files, then pasted it to the Backup library, so after that if i am right the folder that was copied was already cached. but when i re-pasted it again on the Backup library, you will not notice any performance increase, the time elapsed from the first copying are the same on the second, third and so-on copying (note i have still a lot of extra free cache available).
but when you do it directly to the directory on the disk without using the library folder, the first copy will take some time, then second, third and so-on, will copy it instantly without waiting. the copied folder will just re-pasted instantly.
Any recommendation about this?
Thank you
Here is the scenario:
I have two libraries, we will named it File and Backup, they are pointed to a different folder on the disk where read/write cache is enabled.
now from the File library i copy a folder containing a mixed files, then pasted it to the Backup library, so after that if i am right the folder that was copied was already cached. but when i re-pasted it again on the Backup library, you will not notice any performance increase, the time elapsed from the first copying are the same on the second, third and so-on copying (note i have still a lot of extra free cache available).
but when you do it directly to the directory on the disk without using the library folder, the first copy will take some time, then second, third and so-on, will copy it instantly without waiting. the copied folder will just re-pasted instantly.
Any recommendation about this?
Thank you
Re: defer-write
We made some tests as you described, however, we don't see the issue you reported. Can you check the statistics data "Total Read" in the PrimoCache main dialog when you copy files from the library? Will this value increases?
And what is your Windows OS?
And what is your Windows OS?
Re: defer-write
I created a sample video about the issue, here is the link https://youtu.be/vrCEAVbEnSs
here is my quick specs:
windows 7 x64
8gb of ram
I am using 2gb cache for the OS drive and 512mb for data drive(that is the drive F: on the video)
and my libraries is pointing to drive F:
Thanks
here is my quick specs:
windows 7 x64
8gb of ram
I am using 2gb cache for the OS drive and 512mb for data drive(that is the drive F: on the video)
and my libraries is pointing to drive F:
Thanks
Re: defer-write
I watched the video and don't find any problem. Whether you copy files from library/native folder to another library/native folder, I see the cache is working. The following data is get from the PrimoCache statistics after each operation in your video:
Total Read(MB): 64.08 211.02 261.54 310.59 361 509.73 609.48
Cached Read(MB): ~0 147.12 197.69 246.69 297 445.84 545.59
You can see that only 64MB data which is from source files was read from the disk and next reads were from the cache.
(In each column, Total Read - Cached Read = ~64MB)
You may ask why the total read amount in PrimoCache statistics does not equal to the amount you copied. This is because Windows has its own cache system (File-level caching) and some reads were directly from Windows caching instead of requesting data from the disk.
Total Read(MB): 64.08 211.02 261.54 310.59 361 509.73 609.48
Cached Read(MB): ~0 147.12 197.69 246.69 297 445.84 545.59
You can see that only 64MB data which is from source files was read from the disk and next reads were from the cache.
(In each column, Total Read - Cached Read = ~64MB)
You may ask why the total read amount in PrimoCache statistics does not equal to the amount you copied. This is because Windows has its own cache system (File-level caching) and some reads were directly from Windows caching instead of requesting data from the disk.
Re: defer-write
thanks for the reply and providing how to calculate it, and good to know that its really working.
by the way, had you notice the performance difference on copying and deleting from library to library and the real drive to drive?
I also tried now from the drive to library and from library to drive, and got same result.
as fa as know PrimoCache will give a performances boost, because the file was read/write on the ram as long as you have enough ram for the cache? am i right?
thanks
by the way, had you notice the performance difference on copying and deleting from library to library and the real drive to drive?
I also tried now from the drive to library and from library to drive, and got same result.
as fa as know PrimoCache will give a performances boost, because the file was read/write on the ram as long as you have enough ram for the cache? am i right?
thanks
Re: defer-write
As your libraries are pointing to F drive which are cached by PrimoCache, there shall be no significant performance difference on copying and deleting between them.
Re: defer-write
I was got little lost there...
yeah, that is the reason why i thought first PrimoCache is not working in the libraries, because the noticeable performance difference. As you see performing copy/paste and deleting in the libraries is slower than doing it on the actual drive.
yeah, that is the reason why i thought first PrimoCache is not working in the libraries, because the noticeable performance difference. As you see performing copy/paste and deleting in the libraries is slower than doing it on the actual drive.
Re: defer-write
I'm sorry for the late reply because we were on public holiday.
I think this performance difference might be caused by Windows internal caching system. You need to avoid the influence of Windows caching system when doing the test. As Windows caching uses free memory to store files, you can flush Windows cache by copying large files (total size > free memory size) between the first and the second copy/paste.
I think this performance difference might be caused by Windows internal caching system. You need to avoid the influence of Windows caching system when doing the test. As Windows caching uses free memory to store files, you can flush Windows cache by copying large files (total size > free memory size) between the first and the second copy/paste.