Hi all, Two questions: 1. Has anyone here used short stroking and what results has it given? 2. Some people seem to make a small partition and leave the rest of the drive empty, but others have suggested using the larger partition for storage. Is this still short stroking? Can Windows be trusted to keep the smaller partition on the fastest parts of the disk?
This is pretty common practice in enterprise storage. You can get typically get the same (or better) performance by using the outer cylinders of fewer, larger drives than using the entire capacity of a greater number of smaller drives, for a significantly lower TCO - power, maintenance, rack space etc. And you get free storage out of it, kind of. For example speccing out an array with 90x 450GB drives and just using the first 240GB of each (for around 20TB) for high performance storage is likely to be faster than 150x 146GB (for the same 20TB) drives and cost the same. The extra 20TB of unused storage is essentially "free", and you can get away with using this as low-performance storage without affecting the high performance storage. The same principal applies to a single drive, somewhat, but is really meant high I/O throughput, which you're not really going to see on a desktop. Technically you'll get all the advantages of short stroking (even if these advantages on a desktop might be tiny) even when using the rest of the drive, so long as you ensure there's very low I/O activity on the "extra" bit.
Cool, thanks. How does windows decide what goes where on the disk? Does it out the most accessed bits on the fastest sectors of the drive, or put them down chronologically, etc.
When you look in disk management it shows you where on the drive the partition is located, the start is the outer most tracks. Where windows puts data within that partition is down to windows really, defragging should in theory put the most heavily accessed files at the start of the disk. In theory.