An amusing in a the author doesn’t realize at a plonker he looks/is by writing it came up on Slashdot last night.
Alex Payne a senior techie at twitter (a site notorious for crashing more often that James Hunt) has written an article saying essentially that the internet “was built wrong” Who Knew 🙂
Well Alex – yeh back in the day OSI was far superior to this crappy knocked out in a hurry kludge that is the underlying TCP/IP protocols.
Back in the day I used to work on “Proper” OSI networking that is X.400 Mail and X.500 directories for BT. Now OSI was “better” than TCP/IP and X.400 pisses all over SMTP but…
OSI and related standards where hard sums (even now Cisco gets basic facts about OSI networking dead wrong).
At BT we had a full sized cupboard that was full of one OSI standard – it was also more expensive and different OSI stacks had problem with interworking. ICL (the now defunct mainframe company) Famously managed to ignore a key part of one X.400 standard that said “counter MUST! Start from 1) and started it from 0 and i’m sure you can see the problem that is waiting to bite there.
Sprint ignored parts of the standard and had the cheek to say well write a special case just for us.
In contrast the “Internet protocols” where relatively cheap and people with the skills where much more avaible OSI networking was normally the province of the Big PTT’s and Very Big companies like IBM.
As a commenter SatanicPuppy on slashdot says
“It’s very easy to sit back today and say, "Wow it could have been so much better!" But that is armchair crap at the best of times…I’d sneer if Vint Cerf said it. Coming from someone who demonstrably can’t do better, and can’t even be bothered to champion a specific alternative…That’s as pointless and lacking in content as most of the crap that comes through his crappily coded service.”
I see a lot high level technical people at start ups who display a distinct lack of technical chops – and self appointed “Geeks” like Stephen Fry” can make howlers like saying a LAN is an earlier form of an Intranet and get away with It. And the use of MYSQL buy people at companies like Yahoo! and Craigslist have caused a few smiles by those of us used to working with “Proper” Databases – when it turns out that when you start to scale mysql runs into some interesting brick walls
Oh and btw I used to be 3rd line support for international X.400 interconnects and I wrote a large chunk of BT’s X.400 billing system – in FORTRAN of course 🙂 so I know what I am talking about unlike Mr Payne – though I stopped quoteing my phone numbers in Form 1 Varient 3 in My CV a while back.
(oh and if anyone form Cisco reads this tell John Cambers to stay behind tonight and write out MIDI is not a presentation layer protocol 500 times on the white board)
Haunting Thunder - Maurice Walshe's Blog 
Was this written post-pub, in a rash fit of pique? The grammar, punctuation, and spelling wuold seem to suggest just that.
… I’m a pedant – it’s a calling…!
Well i am a dyslexic and if your going to critisise my spelling you could do with spelling “wuold” correctly.
I also spent serveral years as 3rd line support for OSI networking – and get a bit pissed of when some of the web2.0 types who lack the chops start talking about things they know very little about.
the sad thing is twitter has attracted a lot of VC funding and seems to have senior people who dont know how little they know in positions of power.
W00T!
Yes im not sure if the OSI protocols could have scaled that well or that you would have as vibrant an internet if it was an OSI based system run as a walled garden by PTT’s
That used to hack those of us off in BT/DIALCOM/PRESTEL that teh marketing guys didnt get the best way to grow a big business was not to have very expensive walled gardens but to open it up with flatrate products.
The PG S that was involed in pushing the intyernet always lamented that he could not get BT senior management to buy Demon back in the day
Well, to be honest, I suppose we should have picked either variable length IP addresses or 128 bit but we didn’t. And we bound the TCP/UDP endpoint identifiers very strongly to the IP address which resulted in less flexibility for multi-homing and mobility. Nor did we make better (generalized) use of broadcast media with protocols that take advantage of such media to deliver the same transmission to multiple recipients (multicast is a weaker, less efficient alternative).
On the other hand, the system has scaled by about 6 orders of magnitude over the last 25 years and I think that’s not a bad record.