I did some test with my newly added SSD cache drive but there's no difference between Cached shares write and non cached shares write.

 

Here's my setup and my test so far

 

Asus P8H61M_LX_PLUS_R20

Pentium G630

8GB ram

kingston v300 ssd cache drives

1 3 TB Toshiba Parity disk

1 3 TB Toshiba Data disk

 

The test I did was to create 2 shares

1 using cache and another without using cache.

 

I did a copy a 4GB file from my computer (ssd drive with 1Gbs network) to each shares (not at the same time) and the shares with the Cache was not faster than the normal shares!

I tried with multiple files (6.59GB and 1719 pictures) and the result was the same cached partition is not faster than the non cached Partition.

 

So how to test that the cache is working correctly and that is it faster?

Because right now for me the cahe drive is not worth it!

 

Thanks

 

Daniel

 

What kind of speeds were you getting?

I would try something larger. Those are pretty small files. I would try a larger Blu Ray file or something with a bit more sustained transfer time and you should see a fairly large difference. But also keep in mind your not going to transfer anymore than gigabit speeds anyway since that should be your bottleneck to the cache.

What kind of speeds were you getting?

 

What an important missing Information! 

 

For the 4GB single file I was getting around 100 to 120 MB/s

For the multiples files test I was getting around 90 to 100MB/s

 

Note that i'm still with the trial version of unraid.

Right now I have only plex docker running on SSD.

 

My main usage will be to stores files/documents and stream movies in the house.

 

I would try something larger. Those are pretty small files. I would try a larger Blu Ray file or something with a bit more sustained transfer time and you should see a fairly large difference. But also keep in mind your not going to transfer anymore than gigabit speeds anyway since that should be your bottleneck to the cache.

 

I will try it but I know that the bottleneck is the network.

So if my Regular drive are able to substained the gigabit speed the SSD cache is worthless?

 

Thanks

 

Daniel

So if my Regular drive are able to substained the gigabit speed the SSD cache is worthless?

Not worthless, it should make your plex much more responsive than it would be on an array drive.

No that's not what I meant. I was saying if you try a large sustained transfer, you will see it slow down writing directly to the array and you will see your cache drive sustain those gigabit speeds.

 

But as to the ssd being worthless, I use a 1TB spinner for my cache. I do this simply because I use unRAID as a NAS and only a NAS so all of my files are large media, not running VMs and dockers off cache where you might see a benefit.

What kind of speeds were you getting?

 

What an important missing Information! 

 

For the 4GB single file I was getting around 100 to 120 MB/s

For the multiples files test I was getting around 90 to 100MB/s

Is that for both tests? If so then you must have been using cache both times without realizing it. Writing to a parity-protected disk is noticeably slower than 100.

Not just noticeably slower. Should have said significantly slower.

 

Most likely if you are performing the test without cache then you are just experiencing memory caching, which will buffer at network speed until it fills up.

Not just noticeably slower. Should have said significantly slower.

 

Most likely if you are performing the test without cache then you are just experiencing memory caching, which will buffer at network speed until it fills up.

Please note OP is using only 2 array drives, 1 data and 1 parity, which negates the array parity write penalty. I would expect any decently modern system with only 2 array drives to saturate a GB link in many cases.

I did a test with an 37GB blue ray file and got the same throuput around 108 MB/s

Tonight I will add another drive to do more test!

 

Thanks

 

I did a test with an 37GB blue ray file and got the same throuput around 108 MB/s

Tonight I will add another drive to do more test!

 

Thanks

note that the case of having only 1 data drive is a special case that gives you much higher throughput writing to the array than is the case with 2 or more drives.  If you want to do genuine performance testing of writing to the array you need to add another data drive (Even if it is only a small one).

I did a test with an 37GB blue ray file and got the same throuput around 108 MB/s

Tonight I will add another drive to do more test!

 

Thanks

 

As itimpi pointed out when you've got exactly one parity disk and one data drive, it's a special case that is basically RAID1 (more or less...) when you add that second data drive the performance will take a pretty bit hit since the parity calculation becomes more complex, and the read requirements to write parity grow a lot.

 

Usually people report 30-60 MB/s for writes to a parity protected array with more then one data drive and no cache drive.

 

Beyond that also pointed out earlier is the more important point, that having a cache drive allows you to put apps that do a lot of reads and writes on non-array disk allowing you to keep your parity + a data disk from spinning all the time.

Another Question and I already ask it without reply!

https://lime-technology.com/forum/index.php?topic=47701.0

 

With my setup pentium G630 do I will see  day to day performance improvement if I upgrade the CPU?

 

Or it's not worth it

 

Thanks

 

Daniel

Anyone have a clue on the cpu upgrade effect?

 

Thanks

I had an E8500 with 2GB Ram and upgraded to i3 4170 and 4 GB Ram. I use unraid only as backup NAS and i don´t see much better perfomance. But i don´t use any docker or VMs. Bottleneck is the network.

 

Ronny