Sector Skew gives the best performance with defraggler

[PaK PoK]

Undoubtedly defraggler does an excellent job while putting the files in a sequence sector on a disk whereas it proves to be the exact opposite while pulling the data off the sequential sector, just because of the time it takes to transfer the data it does not give the appropriate expected performance. But most of the times the given “sector skew” i.e. not sequential sectors depending upon the disk and the system somehow give you the best and faster performance.

Have you ever thought instead getting the minimum fragmentation you could have Defraggler lay the data down for maximum performance as the defraggler is capable of analyzing disk performance?

But the only thing goes around my mind is that does the hard drive really capable of running faster than the CPU can process?

[Essien]

Not completely but yeah I would partially agree with you. I have noticed that sequential non fragmented drive is lot faster than the roughly fragmented drive. And also being faster it also moves all the system files together than the user files.

This would eliminate windows possibly having to locate a system file in one sector then later on it will have to jump across multi-GB user videos, files etc to find another sector’s next system file. And if the pattern is maximize then you can easily put the system files together and then the user files instead of just moving files that is to be sequential. By having such things the defraggler would definitely take the system one step ahead.

[GaganjyotTechie]

But the only thing goes around my mind is that does the hard drive really capable of running faster than the CPU can process?

Normally the processor can handle the hard drive input very well and smoothly unless it’s a faster and best SSD or something like that. The slowdown on the HDD occurs just because of the files get spread out all the way. The HDD has actually a limited read speed, which normally however have an impact by the file placement. The hard drive could be bought back by moving all the free space then the user files and the system files together.

[DanielaA]

I usually think that this thing was possible and easy to implement with the DOS. Not sure with the numbers but I guess there used to be some 26 sectors in each track with an identities in the sequence like 1,9,17,2,10 etc.

The CPU used to process the data which was held in one sector during the time and therefore it required some 3 rotations in order to process the complete track. Generally if it takes a long time to process one sector then definitely it was bound to miss the next ID and therefore this resulted in waiting for a complete rotation for the next attempt. Now a days the modern CPUs are too quick and fast and also the modern systems has lots of silicon support and not just this also the drive themselves had megabytes of RAM cache.

[GALIENA]

Yeah I completely agree with the above post. Now a days the modern drives most probably will have a diminishing feedbacks and returns on this as they all generally includes a Cache buffer that can be ranged from some 512 KB to 32 KB and can be even more. Also in addition to that the cache temporarily stores data which are needed for the larger drives with the faster rotation rates. So therefore due to this reason it is most of the times faster than the sector skew.

[QUENBY]

Everyone knows that the HDD drives are capable of a huge and wide range of speed; actually they all have a maximum range of throughput and a maximum RPM speed. Actually the outer surface of the drive is supposed making a larger circle as compared to that the inner tracks at the same RPM. Due to this a 15,000 RPM Seagate drive normally have the fastest data throughput on the outer edge of the drive as compared to the central smaller rings. Having the faster part unused it would never make any sense to locate the data at the slower part of the drive.

[Hollyn]

Allow me to share my experience. If you look into the past I initially started with AT MFM type PATA disk driver, if you all know it was a product by Steve Gibso. It used to do all kind of thing to the lower level formatting i.e. physical arrangement of data on hard disk drive surface. I would really suspect that the time taken by the disk head in order to physically reach the next sector is actually perfect for the system to get set ready for the “read”, and then that sector is somehow now automatically handled by the low level hardware things.

The disk benchmarking that provides the measurements are likely to come from those processes certainly. I personally feel that it would be real difficulty in having the expectation for the need for the certain data to an optimized skewed sector. It somehow theoretically sounds possible but don’t know what exactly would be the scenario when done practically.

[Leanne]

Let me tell you that I seriously do not see any requirement of the skew sector. And even if there would have been any such need then it would have been met by the hard drive designers and manufacturers instead of the media have any data. I am somehow starting to expect that the standard defrag will be free of data loss, if the partition image backup is ready in case of any requirement.