The answer is that he's flat-out wrong, and by several orders of magnitude.
The data structure involved (for logical partitions inside an extended partition) is a linked list, meaning that each partition definition contains a pointer to the next one in the list. A linked-list data structure can therefore extend very far. In the case of MBR, the 32-bit limit on sector pointers means you can have a maximum of 2^32 (4,294,967,296) sectors. Each logical partition's definition takes one sector, and if you wanted to max out the number of partitions, you'd make them all one sector, with no gaps -- just partition definition followed by partition followed by partition definition, and so on. (Many tools create extra gaps, but this gap-less arrangement would yield the theoretical maximum number of partitions.) Thus, you'd need two sectors per partition, so you'd have half as many partitions as sectors -- that is, 2^31 (2,147,483,648) partitions on a disk that's 2^32 sectors in size. The MBR itself takes one sector, though, and because they're allocated in pairs, this means taking two out of the count, so the theoretical maximum would be 2,147,483,646.
I have no idea what anybody would want with over two billion single-sector partitions. Also, it's likely that you'd hit OS limits well before that value. Certainly you couldn't map more than a couple dozen partitions to drive letters in Windows, although it's my understanding there are other ways of mounting partitions in modern versions of Windows. Linux has a 16-partition limit in its kernel, although modern distributions implement workarounds that raise the limit -- but I don't know how high, offhand.
Speaking experimentally, I've done tests that have created over 100 logical partitions on a single MBR disk. I didn't do much with my test disk, though, so I can't say how well it would have worked in any given situation or with any particular OS. It was accessible, though, at least in Linux.
Incidentally, I'm the author of the GPT fdisk (gdisk
) partitioning software. Although this is primarily a GPT tool, it does convert to and from MBR format, so I needed to learn MBR data structures pretty well when writing it, hence my knowledge of the subject matter.
As the video to which you link is a training video for a CompTIA exam, there is the practical question of how you answer a question you see on such an exam. The exam's authors are not infallible, so it could be you'll be best served by answering incorrectly. I'd suggest basing the answer you choose on the ones offered (if it's multiple choice), and pick the one that seems closest to being correct.
One more point: MBR is passe. Since the release of Windows 8, the industry has shifted hard away from MBR and to GPT. GPT doesn't use extended or logical partitions, and with their absence, the modifier "primary" becomes meaningless (although it's still used in some partitioning tools). Adhering strictly to the GPT standards creates a disk capable of holding at least 128 partitions. That's a lower limit; partitioning tools are free to create larger partition tables, but few do so in practice. Of course, not all those partition "slots" need to be filled; you can have a GPT disk that holds just one big partition. Depending on when the CompTIA exam you take was last updated and what its subject matter is, you might or might not see any questions on GPT.
1I would recommend commenting on the video – SeanClt – 2016-02-24T01:06:47.287
Thanks for the reply. I have done that, but no reply. He is a senior professor as you can see here: https://www.linkedin.com/in/lowell-vanderpool-a442085 Have you seen the video at 8.50. according to you is it 23+3 or 32.
– Neeraj – 2016-02-24T01:16:14.570Can you provide some background research for this topic, as in, what your understanding is based upon? Documentation suggesting either 32 or 23+3? – jhaagsma – 2016-02-29T19:14:34.303