In message <199610141906.PAA23721@panix.com>, Dana Hudes writes:
Point may have been obscured. You add more facts (thanks!) the question is what the $10 million represented -- did it include the money IBM spent but did not charge the NSF?
It represented what NSF spent. I think AT&T bid $22M/year in 1987 for a T1 network, though I didn't read that anywhere. The OIG report describes the non-competitive nature of some of the alternate requests.
As for the NSS's being still in place but disconnected, I thought the lease was up a year ago. The NSS's certainly are not cost effective. One of the problems that they faced in the versions subsequent to those you had at ANS was running out of backplane bandwidth for WAN-LAN (and LAN-LAN) transfers. Another of course was their proprietary nature. The price tag doesn't surprise me, Rs/6ks are expensive. But the sampling feature would be helpful in the SYN attack situation.
The NSFNET was over in May 1995. IBM and ANS contracted for ANS to pay IBM to support the routers for another year to allow time to find a replacement and get them deployed. Things ran way behind schedule so after June the NSS were no longer supported but still in use for about 2 months.
Dana
Curtis