Mr. Vinod

To get some information is a process of two phases or steps.

1) To Look For the information

2) To Get the information, that could include some filter and analysis sub-phases

If you want to recover contiguous data you just make a Look For information phase a then a large get or recovery phase.

If you would like to Access dispersed information you must make many differents Look For phases which mean more time.

Aditionally you normally Access information by blocks no by words, that is common in HD because it is organized by blocks, contiguous information could be in the same block than the previous information that was recovered previously, that means less time To Look for them.

Even with RAM memory where its organization is viewed as a random access, where any memory Word could be accessed in the same time, the true could be a Little bit different.  Current computer architecture always uses some kind of cache memory and also virtual memory, where page or segment are used more or less in the same way as blocks in a HD. To move information from the RAM memory to de CPU, because of these features, contiguous information could be accessed in the same cache memory block, which mean that could be accessed in less time, than other disperse information, that have to be looked for on other block of RAM that have to be carried from the RAM to the cache memory, or that means a page fault in a virtual memory scheme.

I hope that this information could be useful for you.

Dear Ricardo

Thank you for your inputs, these look valuable.

Now as SSD is similar to RAM memory so similar explanation holds for SSD, I believe.

My English is a bit too weak to explain everything well. In order to answer your question correctly, you must first understand the physics and mechanics of your hard drive.

When you think about data, you usually think of files in directories on partitions. As long as you are a user, it suffices. But the disk does not understand it. Disc understands the language of the electromagnetic field. Therefore, in order for the disc to show you what you see as a user, the electromagnetic field language is translated into the language of images and documents. This is not easy and takes place in several stages.

Let's start with files - so it seems to me simpler.

You probably already know the concept of the sector and LBA addresses. Sector 0 (MBR) contains key information, including information about partitions on the disk. Partitions addresses are expressed in LBAs. Each partition has its logical structure. Directories and files that are also addressed directly or indirectly through LBA addresses. The data you are thinking of are probably located in consecutive or random LBA addresses. You probably know the defragmentation procedure that links the scattered files to the continuous strings of LBA addresses. This is to improve system performance and usually does it, but not always. I assume that you already know that this has prompted you to consider this topic.

Do you know how LBA sectors are distributed on disk? You probably know that one or more plates are inside the disc. Each of the plates has two surfaces. Each surface is read by one head. It is not always the case - but for simplicity, let's assume that there are no exceptions. We have many concentric tracks on each surface and many sectors on each track. Do you think that on every surface we have as many tracks, and on each track as many sectors?

I'm sure you can calculate the circumference of the circle. If you take a track radius close to the disk axis of 20 millimeters and at the edge of the plate equals 45 millimeters, you will quickly guess that to increase the density of the record, it is useful to differentiate the number of sectors in the track. So do manufacturers. Tracks are grouped into zones with different sectors per track.

If you look at the read rate graphs for each of the LBA sectors, you are likely to notice significant changes in read speed at transitions between successive zones. I suggest that you make this observation on a new disk to reduce the impact of deviations from disk consumption.

Now a few words about the number of tracks per plate. The technology of making magnetic plates does not allow for high repetition of their parameters. Therefore, the number of tracks on the plate surface may also vary. Zone tables in the latest disks are very extensive as they must contain information about the number of tracks and sectors per track in each zone for each surface.

 In addition, we have defect tables. You've probably heard about spare sectors and remaping of bad sectors. Usually the disk has two defect tables. One is still in the factory and allows you to bypass the factory defects of the surface of the plate. The second one is created during operation and allows to hide defective sectors by assigning corresponding LBA addresses to the physical addresses of the reserve sectors.

Now that you know it, it is not a surprise to you that LBA addresses are not real disk addresses for real sectors. It's just such a universal translation layer that allows different file systems to communicate with different disk models. Or RAID arrays. There are also LBA addresses. A known from the past, the physical addressing of the CHS sectors still exists. The above-mentioned zone tables and defect tables are part of a system that allows the translation of LBA addresses to physical addresses and vice versa. LBA addresses are generally assigned to successive sectors on subsequent paths. But each path ends once ... you have to go to the next or switch to another head reading another surface ... or find the path to the backup sectors to bypass the defect ... How do you see the continuity of data in address LBA easily breaks when we go to physical address. The key to answering your question is understanding the physical layer of the disk. If you need to know more or any part of my answer is unclear to you - ask.

About the SSD i will write later.

To greatly oversimplify, a HDD is going to be significantly slower at data recovery than an SSD because of the mechanical element.  How and where data is written to the disk as a contiguous cluster (block) or as a fragmented file (non-contiguous clusters) depends upon the disk file system.  Windows NTFS, for example, utilizes the $MFT system metadata file to store runlists or pointers to data locations on disk.  Linux EXT4, for example (along with other Unix-derivatives), use a Superblock with corresponding inodes, etc.  I know this is somewhat vague, but an understanding of file systems might be useful in your research.

I'm going back to the subject. Today a few words about the SSD. Here you should not have a noticeable difference between continuous and distributed data. SSDs (and all other media based on Flash-NAND chips) always dissipate data. That is their principle of work. Dispersion of information between different circuits allows the entire media to accelerate. The SSD on the external interface also uses LBA addresses, but the Flash-NAND chip itself does not understand the term "sector." The basic unit for Flash-NAND is the page. The page size is related to the size of the data register and is equivalent to 1 to 32 sectors. At least in my practice I have not met bigger pages, but I expect that in the future they will appear.

But page size in bytes is not a simple multiplication result for example 32x512. The page is always bigger and contains ECC codes, various markers, sometimes other information necessary for proper media operation, and sometimes garbage, eg no meaningful strings 0xFF.

The controller writes the data to the pages in different layouts and then reads them from those layouts. You can be pretty sure that when you save a large enough file, parts of it will be in all Flash-NAND chips of storage media.

I have already mentioned that semiconductor media like SSDs also use LBA addresses. In their case, it is even better to see that the LBA address is not a physical address, but a translation layer between the physical addresses of the media and the file system. While LBA addresses on hard disks are quite closely related to physical addresses, SSDs are dynamically assigned and vary with each write operation. This involves the physics of writing in NANDs, which does not allow you to edit data, but requires erasing an entire block that counts multiple pages and then saving to a page by the page. So when you edit a file in an LBA address by changing one character, the SSD must perform a series of operations:

In simplification:

- reads the data block into the buffer

- means this block, as empty, intended to be erased,

- changes the contents of the block according to the command received from the computer,

- writes data to another physical block, and from the file system point of view these data still remain in the same LBA.

- the released block is deleted and you can save another data in it.

In fact, the process is much more complex. The algorithm must take into account requirements related to the speed and performance of the media and strive for a uniform consumption of the chips (wear-leveling). Because erase and write operations are destructive to Flash-NAND memory.

This way of writing and reading data makes you should not see the difference between continuous and distributed data. From a media standpoint, there is no such thing as continuous data. This does not affect its performance. Therefore, the SSD does not defragment.

Likewise, you should not see the difference in RAM. If you see it, you can actually search for cache or other data buffers (are you sure the requested data actually reads from the media, whether you get them from the buffer?), or in a system failure that may require additional time to correct errors or re-read.

Answers to your query do not look in the logical structure of file systems. The physical media do not understand of these structures.