Date: Sun, 24 Sep 1995 09:26:21 Est
Reply-To: "Daniel Pimienta " <[email protected]>
From: "Daniel Pimienta " <[email protected]>
To: "Laurence I. Press" <[email protected]>
Cc: [email protected], [email protected],
[email protected]
Subject: Re: Redes en el Caribe
Larry,

>> Both networks CUNET and REDID are so far UUCP which means there
>> is no WWW browsing in the traffic pattern. Only e-mail (including...
>Archie speaks of IP connectivity in many nations. While some of that
>may be LANs or intra-national, Larry Landwber's table lists Barabdos,
>Jamaica, and Bermuda as IP-connected. (Maybe others too, I just
>looked quickly). Are there other nets beside CUnet and REDID?
Jamaica is TCP-IP indeed but was not included in the CUNET traffic table. Saying that Jamnet is part of CUNET or not is a semantic question in which I would not like to be part: I just know that this network (more exactly the UWI) received consistent support from various sources (Canada and European Union being the most) and the architecture of the net have nothing to do with the one of CUNET (star topology to UPR). Beside CUNET, REDID, REHRED, Jamnet, CENIAI, and Puerto Rico's nets there are other initiatives in terms of BBS, some having numbers of users in the order of magnitude or superior to CUNET.

As for IP connectivity, almost all the Caribbean is now connected by commercial providers and Larry's table should reflect it at one point of time. My guesstimates is that the number of commercial provider users (including researchers opting for that type of access) is far above the number of research network access except in Cuba (where there is NO commercial IP). I am not sure about what is the situation in Puerto Rico and I suspect that the commercial IP market segment are still below the academic since the commercial IP open services only few weeks ago.

(NOTE: This situation of commercial IPs is

  1. good news: global growth and opportunity,
  2. diagnostic: weakness of academic and NGOs networks
  3. preocupation: how will the non profit sector fill its room in the Caribbean infostructure?)

DIFFEREMCE OF FIGURES
As for your effort to understand the differences of figure, let me tell you more. You may be right that some traffic behaviour differences may explain part of the difference, but that would be the minor part only. Between 130 and 2000, my estimate would be that the accounting of inactive users explain the difference of magnitude (1000%) and maybe a lower propoensity to use in CUNET (due to volume charge and blocking factors) may explain a difference of less than 100%.

Let me back-up my estimate with some data. My personal knowledge of CUNET is limited to the direct case of Dominican Rep. and Barbados, and I have indirect data from Santa Lucia. Within those three cases, crosschecking confirm my user estimations. I DO NOT know if these three cases are representative of the whole CUNET but I my algorithm have demostrated in the three case being correct (by crosschecking).

DOMINICAN REPUBLIC.
In Dominican Republic, the CUNET node is a PC with a single phone line. The same unique line is used both for user connection and for gateway to Puerto Rico. Users create or read there mails dialling up this line. The system automatically calls Puerto Rico to send/receive in a scheduled fashion (2 or 3 times a day, I'm not sure). The system is shut down during the week-end.

Any person familiar with queueing theory could compute the maximum number of users such system can handle with acceptable waiting time. Remember to compute the writing and reading time of messages as part of the connecting time (including the consultation of mails archived).

The operating system is Waffle. It tells the actual user who was the last connected and a command allows to list the registered users with their last login data. The modem is 2400bps. The history of the system shows frequent periods of system down of several days/weeks duration (a small part of them are a consequence of UPR2 failures that unfortunately REDID shares). When a remote user is connected, the system screen reflects exactly the same as the remote user screen.

As for more subjective data, one of the most traffic prone user of REDID is also using CUNET/DO (this person have counterparts in Africa where the doble bam format "pimienta!daniel" is not interpreted by the mailer so the person cannot receive from them with REDID). This person is probably accountable for a large chunk of the CUNET/DO 3.5 MBytes/month traffic. This person reports few different Id shown as "last login" since the 2 years of use. Besides, there is the system administrator Flavio Moncion and few people from the university (PUCMM). Flavio says and writes he got 140 users. I have no reason to doubt there is 140 people effectively registered in the system but I suspect there is less than 10 using it. We have repeteadly asked him to publish the user list since 1992, so that to help communication within DR. We never got anything, and REDID unformal surveys have not allowed to identify more than 5 users answering mails there. If one susbscribes the listserv SDOMINGO and ask for REVIEW, only 3 CUNET/DO addresses appears in the list managed by the system administrator. For your information, the campus of PUCMM in Santiago is having 25 users in REDID.

=====>My user figure projection says 7 users. I could be wrong
of 100% (14) but certainly not of 2000% (140).
BARBADOS
FUNREDES was contracted, last November, to connect and train the NGO Caribbean Rights in Barbados. We spent, Pablo Liendo and I, one week there and learnt a lot about this CUNET node, partly thanks to the nice cooperation of the system administrator. It is the same system as in the DR, except it remains connected in week-end and the modem is 9600bps. Our contact was the first connection to CUNET out of the University campus, and, the first NGO connected. The list of registered user was reaching 40 with around 10 last logged within the last 30 days, data coherent with the acceptable level of blocking we experimented during our 4 days of testing.

====>My projection says 17. I could be wrong of 100% (34),
but certainly not of 2000% (340).

SANTA LUCIA
After we completed the report, a user of Santa Lucia network (this CUNET node is operated by an NGO and there is a clear will for user disemination) send me the Waffle user report. I will transmit it to you in the next mail. I discovered my algorithm was online with the data. The report shows 123 registered users, 34 of which have been connected within the last 30 days and 23 within the last 10 days.

=====>My projection says 24. No comment.
I hope this will help you. I expect you will feed me back on any complementary data about Caribbean Networking.

Thanks,
Daniel

PS: If Saul or Archie could publish the CUNET user list together with
the information of when they last logged, the issue would transform from mystery or speculation into facts. .