NO MORE UPDATES TO THIS PAGE PLEASE. SUBMIT ALL FUTURE COMMENTS TO

Evnard's diagram (pg 5)

Doesn't mention upgrade, as distinct from update. Rebuilding a server from Solaris 2.6 to Solaris 9 is more than an update. A lot of people will simply do a rebuild to get to "clean" state, and so there is a logical step from stable to clean via rebuild.

[ Tom's reply: Noted. Status: DONE. ]

footnote on pg 8

Windows NT4 has always had unattended install options (when we did this our manager initially malaprop'd "Unintended installs" which we thought very apropos). See, for example http://www.microsoft.com/technet/archive/winntas/maintain/manntnet/ntnfch11.mspx?mfr=true - it was also possible to do this via network installs using specially created boot floppies (since netboot PCs weren't common) from the NT Resource Kit. Maybe just worth adding "or Windows Unattended Installs" to the end of the footnote, for accuracy.

[ Tom's reply: Yes, AutoLoad^? was a wrapper around WUI which made it useful and easy. Since they now called it RIS and it is well-supported, I've updated the text to just mention RIS. Status: DONE. ]

Use of Dynamic Leases with DHCP (pg 20)

You state that office LANs are better suited for static. Not really true for large sites. A large multi-site environment (eg a bank) with thousands of desktops will use dynamic leases simply because the sheer headache involved in keeping track of MAC addresses as people move or machines are replaced (repairs, replacements, upgrades). It'd be impossible. With firewalls able to do NAT at network borders and proxies to control most external accesses, internal LAN IP addresses could easily be given out from RFC1918 ranges (eg 10.* or 192.168.*) so the corporate LAN has plenty of spare IP addresses to waste.

[ Tom's reply. That who section is now rewritten. Status: DONE ]

Discussion of customers (pg 25)

Is it worth mentioning that large sites might have special SA teams who have a primary role of engineering standard builds and providing third level support to the SAs who deploy and manage the resulting systems? These engineering teams essentially have two levels of customers; the SAs who do the daily work and the end users.

[ Tom's reply: Noted. Not sure if I'm going to be able to work that in. Status: DONE ]