My company is a bio tech facility, concentrating primarily on gene discovery. Our network is a rather large scale unix environment with a significant contingency of macintosh clients.

We have been in a rapid growth period for the past two years, and we now find that in addition to scientists, RAs, and post docs, our executive and administrative staff is increasing exponentially.

This particular user community is comprised exclusively of Win95 clients, and as such, has created a need for our systems department to re-design or "re-think" our NT environment.

Of course, since our NT environment grew from a small test network to a full blown corporate solution in a very short period, we are faced with many problems without having the luxury of starting from scratch with a well planned implementation. We are currently researching how best to facilitate services such as browsing, printing, and file sharing in a WAN environment. Until recently, we were getting by running NetBeui with bridging enabled at our routers; however, now that we have multiple sights, this is no longer a viable option. It is apparent that we need to implement net bios over IP in order to reduce network traffic created by broadcasts. This is a bit more complicated than it seems, as we will have to reconfigure our routers to forward IP broadcast datagrams. In our research we have familiarized ourselves with the technical elements of getting a complete, domain/network-wide browse list, but have not heard the specifics as to what needs to be done, nor have we read any accounts of a large scale NT based company successfully implementing NBT. Yet another concern regarding browsing over IP in a multiple segment network is the inability of Win95 clients to participate in a domain, therefore limiting thief browse list to the local workgroup. It is implied that since each client would receive its browse list from the MasterBrowser in its subnet, assuming that the DomainMasterBrowser is successfully updating the local MasterBrowser with a composite, network-wide list, the Win95 client should be able to take advantage of a full list as long as each workgroup on every segment maintains the name of the domain. After reading the documentation, and fully understanding the technical process by which this all occurs, we still are not confident that this will indeed function properly. We hope to hear from, and perhaps speak with other sys. admins who have successfully implemented NetBios over IP, in order to gain insight from those with first hand experience with this problem.

Yet another problem which we face is how to deal with multiple domains. In planning this we are first challenged with the fact that NT4.0 is not really doing this well, which creates an entirely new subset of problems. If in fact NT4.0 did multiple domains well, we would still need to determine a number of things: do we in fact need more than one domain, how important is the location factor, what domain model best suites our needs, and which works the best? We know that a PDC can handle a substantial amount of log-ins; yet, should we put a BDC at every location for fault tolerance or loss of network connectivity between sights.

We look forward to discussing Cairo and NT5.0 with regard to some of the enterprise issues such as multiple domains and unix integration, as well as native and third party utilities for system maintenance and automated tasks.

David St. Denis
Systems Engineer
Millennium pharmaceuticals Inc.
640 Memorial Drive
Cambridge MA 02139
(617)679-7193
st.denis@mpi.com