RE: Choice of ESP alg. for IPS/IPSec - 3DES-CBC vs. 3DES-CBC-I

["Mukund, Shridhar" <Shridhar_Mukund@adaptec.com>]

Bernard,

> -----Original Message-----
> From: Bernard Aboba [mailto:bernard_aboba@hotmail.com]
> Sent: Wednesday, December 05, 2001 5:14 PM
> To: Shridhar_Mukund@adaptec.com; pkoning@equallogic.com
> Cc: ips@ece.cmu.edu
> Subject: RE: Choice of ESP alg. for IPS/IPSec - 3DES-CBC vs. 
> 3DES-CBC-I
> 
> 
> >When AES-CTR is
> >approved, AES-CTR becomes MUST and 3DES-CBC is demoted to MAY.
> 
> Doing that will create a problem in interoperation between 
> iSCSI HBAs (which 
> might have AES support) and software-only implementations, 
> all of which now 
> support 3DES-CBC. So it seems like 3DES-CBC has to be a MUST.
> 

>>> Yes, if we do not prep appropriatly we are inviting 
>>> trouble down the road. But then, we do agree that it is the
>>> charter of the IPS WG to enable low-cost iSCSI solutions from >>> today
all the way up to 10G.

>>> The IPSec implementation(hence interop) complexity is really 
>>> around yIKEs! Given that we leave the mathematically inclined
>>> folks to invent(and standardize) ESP/AH algorithms, software
>>> implementation of these algorithms is the easy part. 

>>> One of the important motivations behind AES is to simplify
>>> s/w implementation. Since 3DES is compute intensive, 
>>> s/w implementations will transition rapidly. Even today
>>> several implementations taut AES.

>>> On a lighter note, those who resist small s/w changes that 
>>> bring significant value are not in this audience. iSCSI
>>> is about change.

-Shridhar Mukund