Jump to content
Division-M Community
Sign in to follow this  
Davey126

Memory Cache

Recommended Posts

Curious if anyone knows if DriveBender uses its own memory cache or leverages the Windows file cache. I was benchmarking drives and saw some crazy numbers posted for my DB pool ... faster than any mechanical drive and all but the most advanced SSDs. This could well be an artifact of the Windows file cache although I did not observe it on any other drive. The only time I have seen benchmark numbers in that range is when a dedicated memory cache sits in front of the drive (eg: Primocache).

 

More of a curiosity question. Of course, if DB does create its own cached I'd be curious as to its size and whether it can be tweaked via registry settings  ^-^ .

Share this post


Link to post
Share on other sites

Yeah - there is clearly some caching going on whether it be Windows or Drive Bender (see below). Not bad for a pair of 500GB spinners; check out those access times - lol. The cache must be relatively small as evidenced by the trail off at the end of the ATTO run with larger data sets. Still pretty cool to see those numbers, especially on my modest core 2 era rig.

 

I could dig to figure out where the caching is taking place but in the end it really does not matter. I'm getting great performance out of the pool for my usage pattern. Kinda fun to back perception with numbers.

 

BTW - the Drive Bender Manager concurred with the read/write numbers during benchmarking. The read value actually overflowed the allocated space at the bottom of the UI. The developers probably didn't anticipate throughput numbers exceeding 999 MB/s. Also suggests the UI isn't measuring raw IO at the physical disk level.

post-1683-0-06952300-1411093383_thumb.png

post-1683-0-80384800-1411093408_thumb.png

Share this post


Link to post
Share on other sites

Another cache place is the drive itself. How many modern drives have less than a 16MB cache?

 

I will note that there is one thing I do on my pool that runs about 4 times longer on my 3 drive pool as opposed to a native drive. It's the database populate run for my music system (Squeezebox LMS software). This software reads all the tags from my music files and builds a SQLite database. The tags are generally in the 1st 4KB of each song file. For my 19,000 song library it takes about 13 to 15 minutes reading from the pool drive (all music files are dup'd). On my laptop this same run takes 3 to 5 minutes. In both cases the database being built is on a native drive and in the DB case the music files are being accessed from the DB pool.

Share this post


Link to post
Share on other sites

Good thought - both drives have 16MB caches. But the files written for benchmarking are far larger.

 

I too have noticed Drive Bender struggles rapidly accessing large numbers of small files on pooled drives. Some of this is due to OS overhead but it's clearly slower than accessing the same files off native drives. In my case I typically work with only a few files at a time - both small and large (sometimes dozens of GB each). Most involve simple moves/copies with occasional edits. For these types of operations DB shines. 

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
Sign in to follow this  

×
×
  • Create New...