Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!accuvax.nwu.edu!nucsrl!telecom-request From: Joe Stong Newsgroups: comp.dcom.telecom Subject: A Puzzle Message-ID: <3787@accuvax.nwu.edu> Date: 13 Feb 90 04:33:30 GMT Sender: news@accuvax.nwu.edu Organization: UCSF Medical Center Lines: 41 Approved: Telecom@eecs.nwu.edu X-Submissions-To: telecom@eecs.nwu.edu X-Administrivia-To: telecom-request@eecs.nwu.edu X-Telecom-Digest: Volume 10, Issue 95, message 4 of 5 I'm sending this to telecom, because I suspect this sort of data keeping is liable to be used by the phone companies... I was asked recently how much disk space would be needed to keep VISA's bad card list: I was told it had 400 million numbers. It is my understanding that VISA numbers have 12 digits, and 4 check digits. (Is this right?), thus 10^12 or a trillion numbers. Raw storage of these numbers would require 40 bits per number, or 5 TeraBytes. Now, a bitmap of a trillion numbers would take 125 GigaBytes. Since the bad numbers are sparse (about 1 in 2500), on the average, the distance between bad numbers could be kept, on the AVERAGE, as a 12 bit quantity (sometimes more, sometimes less, using a bit-level-escaping technique), so, assuming 12 bits * 4 million gives 600 megabytes, (maybe double that to account for the encoding overhead, and indexing to make the lookups fast). Is this realistic? Does anyone know of any better techniques for storing the numbers? Does the phone company do any compression on the tables of who has what features in an ESS office, or in the systems that they use to do billing accounting? Anyone know of any pointers to material dealing with large number storage problems? Send me mail please, or post and send mail, and I'll post a summary of responses. I have trouble keeping up with the volume of NetNews. P.S.: I didn't promise a summary about the X.25 encapsulation question, but many replies suggested that vendors are typically using HDLC packets without an acknowledgement mecanism to carry TCP-IP packets for transmission on T1 lines. I have a friend who "thinks the whole world" is doing what he is doing, which is systems that encapsulate TCP-IP in X.25 packets on T1's for a mostly DOD network. It appears that he doesn't have the big picture. Joe Stong jst@cca.ucsf.edu