Hello, I found powershel script that show how much filesystem memory min and max you have on windows, there is also link that can change that. as I understand this is how much mft windows cache. this amount directly influence how fast system search files on hdd.
Script is open source no need to run any unknowns exe.
pinvoke - c# get system file cache size - Stack Overflow
So any precisions from experience windows users?
We don’t want to reduce metadata caching, so probably not the solution for storj.
this is not about reducing it, we can sett it bigger, also this information gives us that ntfs mft can also be cached and work faster.
Isn’t using all available RAM for caching the default? If so using those functions makes sense only to reduce.
my pc use only up to 50% of ram for cache, if ram is needed it reduce it itselft.
So bigger ram fit more mft
before was told that only linux cache metadata, but now we see windows make the same
This is my windows server. All RAM used for something, most of the cache is metadata. I don’t think there is much to gain with tuning.
1 Like
not a lot users have windows server, also you have 50gb of ram? not a lot people have this amount also.
RAM is 48GB which is low in my opinion for running multiple storagenodes but I use what I have…
I mostly have 24-32 gb of ram, now i think about plan haw to upgrade.
In my opinion ZFS might be the best for such system. This sounds curious because ZFS is known for needing much RAM. But when having metadata and database on SSD all remaining HDD access is randow and needs no cache. I am just testing such setup.