NTFS Cluster Size for Video Hard Disk

[cweb]

Hi,
I'm trying to see if it is worth using 64k cluster sizes in NTFS on a hard disk used for video processing.

Some sites give the advice that it makes no difference to the default cluster size.

Does anyone have practical experience?

Are there known issues or problems (E.g file corruption) with XP SP2 using 64k cluster sizes on a really large hard disk e.g 500GB? And if it is externally connected via USB2?

[foxyshadis]

Nah, SP2 won't corrupt at any cluster size. All you lose is compression support, and that's pointless on a huge temp drive anyway. You won't really get more performance, though, mostly just less fragmentation and less MFT overhead. ext3 would be better from a performance standpoint, except with external drives performance is well reduced from the maximum anyway, so it makes no big difference. Maybe compare read/write speed with fat and ntfs over your usb, just to find out if the gains of fat are noticeable externally.

The high and low extremes are mostly meant for servers with very specific needs.

[cweb]

Thanks! I did my own simple test, I ran tmpgenc to process a video stored on the external hard disk. I did it again after reformatting, so I could compare and any speed benefits of the use of ntfs 64k cluster size were negligible.

I'm going back to the default. That means a reformat.

[Atamido]

As long as the space usage stays below 50%, you shouldn't have any problem with framentation. If your drive has >90% usage and/or you mix large and small files, then fragmentation is likely to occur and a larger cluster size will help reduce fragmentation.

[cweb]

But since I did encounter a serious problem last weekend (a corrupted directory) when I tried to use the 64k clusters, I decided the default 4k cluster size to be safer for me at least. I did find a posting on some website with a very similar problem with 64k clusters, so it wasn't just me.