Scaling Factor

Question

Header field need scaling factor, , when it’s size is small, that is ok

But why fragment offset field also need scaling factor?

https://gateoverflow.in/1548/gate2013-37

Comments

first tell me why header length is using scaling factor here?i.e. 10*4

And then tell me in final calculation they just used 10 as header length ,why? i.e. 10*10+9*20

why removing scaling factor from header length here?

---

1) Mam because scaling factor of IP header = 4

 we know that MAX header size of IP header = 60 Bytes

 and MIN header size of  IP header   = 20 Bytes

 Now ,In IP header HL = 4 bits  = 1 1 1 1 = 15

 So how we did the representation for 20 B or 60 B ???

 that's Y we going to use a scaling factor of 4

 60 / 15 = 4

 whenever in the question given HL = X

you multiply it by 4  = 4X 

 

2 )  tell me in final calculation they just used 10 as header length ,why? i.e. 10*10+9*20

 it's given in the question that offset value is  100  means the IP datagram of size 1500B  is   going to fragmented by the  router into 200 B have FO  = 100

that's Y the ist fragment IP datagram having the FO = 100

                     2nd fragment IP datagram having the FO = 100 + 160/ 8  = 120

                    3rd Fragment IP datagram having the FO = 120 + 160 / 8 = 140

                  ..goes on

therefore offset value  last  fragment  = 100 + 9*20

---

@Magma

ok, I got it, range of header between 20 to 60 B every time

then why fragment offset need scaling? https://gateoverflow.in/1548/gate2013-37

Answer 1

Maximum packet size = $2^{16}$-1 bytes = 65535 bytes

Minimum header length = 20 bytes

Thus, Maximum data bytes in payload = 65535 - 20 = 65515 bytes

Now,

Fragment offset = 13 bits $\rightarrow$ 8192 combinations

8192th number should represent 65515th byte

So each number must represent $\left \lceil \frac{65515}{8192} \right \rceil$ = 8 bytes

Thus scaling factor is 8 bytes.

Comments

but for such a big number, we do not use scaling

Such big numbers are the sole reason why we use scaling as 13 bits cannot represent all the 65515 bytes.

 

why 60 byte needed? 

Header contains compulsory 20 bytes to be filled. Apart from this there is 40 bytes optional field which can be used if source routing.

---

Such big numbers are the sole reason why we use scaling as 13 bits cannot represent all the 65515 bytes.

but  question used scaling for 300 B

is there any lower limit?

---

Fragment offset in IP header is defined with a scaling factor of 8 bytes.

Whatever be the value of fragment offset, say p, then 8*p is the starting byte number of the data contained in that fragment.