What's faster - file based I/O or block based?

FAQ, getting help, user experience about PrimoCache
Post Reply
RobF99
Level 8
Level 8
Posts: 130
Joined: Fri Sep 19, 2014 5:14 am

What's faster - file based I/O or block based?

Post by RobF99 »

Hi - I have a question - I once did a test and actually used a second spinning hard drive as L2 cache and it worked. So my question is could I do that and get a faster read performance because L2 is block based or is block based I/O the same speed as file based without necessarily any increased performance? Just interested to know.
User avatar
Support
Support Team
Support Team
Posts: 3731
Joined: Sun Dec 21, 2008 2:42 am

Re: What's faster - file based I/O or block based?

Post by Support »

Well, it's hard to say the performance will be improved or not, or even worse. The advantage of L2 which using another spinning hard drive is that it can reduce the number of original disk's IOs to be processed. Some data can be fetched from L2 instead of original disk, so original disk can process more incoming IOs. The disadvantage of L2 is that the L2 speed overall is lower than L2 disk's native speed because of additional processing time.
RobF99
Level 8
Level 8
Posts: 130
Joined: Fri Sep 19, 2014 5:14 am

Re: What's faster - file based I/O or block based?

Post by RobF99 »

That's an interesting concept in that it almost is in a kind of way like a form of load balancing in that you have I/O's shared across two drives for part of the time before L2 would flush to original disk. Interesting. I might play with that a bit more for fun. If I get time I might do some bench marking tests just to answer the question about the performance difference.
User avatar
Support
Support Team
Support Team
Posts: 3731
Joined: Sun Dec 21, 2008 2:42 am

Re: What's faster - file based I/O or block based?

Post by Support »

yes, kind of like soft-raid..
User avatar
Jaga
Contributor
Contributor
Posts: 694
Joined: Sat Jan 25, 2014 1:11 am

Re: What's faster - file based I/O or block based?

Post by Jaga »

RobF99 wrote:That's an interesting concept in that it almost is in a kind of way like a form of load balancing in that you have I/O's shared across two drives for part of the time before L2 would flush to original disk. Interesting. I might play with that a bit more for fun. If I get time I might do some bench marking tests just to answer the question about the performance difference.
I would never use software for load balancing, based off of years building racks in data center apps with instance mirroring. Hardware is always the way to go.
support wrote:yes, kind of like soft-raid..
Which unfortunately adds overhead on the OS, slowing down operations. It works, but hardware-based RAID is always faster.

SSDs are cheap enough if you get a small one just to use as L2. Even a lightly-used SSD would have enough lifetime to do the job.
Post Reply