I think it’s risky business by Intel. If “8″ doesn’t sell they are out of luck. I wonder how much M$ paid for exclusivity.

http://www.tuxmachines.org/node/58464

Yeah, sure, nobody ever used NT, all those sales and deployments never happened …
Just keep on making up your own reality. Don’t worry about facts, believe instead.

“due to how identical Samba and NT looked to end users”

They didn’t look identical to people with admin accounts.

ch mind you this could be regional as well. Thinking Samba lead developer is in My country.

ch so you are one the odd real NT deployments. Not that common for that time frame.

As I stated you cannot find out what was the deployment unless you add access to the server room around that time due to how identical Samba and NT looked to end users.

More often than not when I question people over this kind of claim its bogus. I had to find one sooner or latter who was using NT really. You were in the minority for that timeframe. Lot were sticking to there Unix with Samba or BSD with Samba.

Its like all shifts there are different groups different numbers. In every statical happening there are edge groups.

Again does not change the fact that was not a really true integrated backbone and it could not be. Of course some business operations do want many small isolate servers not a integrated backbone.

Web of trusts does not make a integrated backbone by the way. Since you don’t have the multi location backups of a proper integrated backbone. That is what cause the problem ch you mostly likely used the wrong term to describe what you had.

No they were not. I was working with the admins all of the time, even had an admin account myself, and I knew what was going on. No, we didn’t even talk about Samba – it was NT, and that was it. Stop clutching for straws!

“So boss would sign off on copies of Windows copies of Unix/BSD not knowing that both were being deployed as one.”

Our bosses would have known, they were no PHBs.

“Basically cheapest so called server network card was a brick.”

Why in the world would you buy the cheapest whatever for a server??? We used mostly 3Com cards at the time, no problems. But then again, we knew what we were doing.

“As you say ch unix was used where appropriate the big important question is where was that.”

Stuff like DB servers (those that weren’t SQL Server for one reason or another), Internet proxy, some special applications etc.

So by that there is a chance that you NT network domain controllers in that time-frame were really Samba sitting on LDAP. This would not be uncommon. Up until year 2000 Microsoft attended all the Plug fests for the SMB and related protocols. So a samba domain controller worked perfectly without risk of having NT insanities. So was quite a popular solution. Even more fun since there is no licensing change with Samba you don’t have to bug the Boss that much to deploy it. So boss would sign off on copies of Windows copies of Unix/BSD not knowing that both were being deployed as one.

This is why its important to check records to make sure what you were using pre 2000. There were a lot of cases of windows servers just being members server in Samba run NT networks in that time frame. So windows not doing the domain control role due to the glitches the Microsoft implementation had.

Ch some of that cheap and nasty network hardware was relabel as good brand hardware as well. –Basically cheapest so called server network card was a brick.– But a general PC version network cards was cheaper without the defect. Yes stuff labelled server that not work. There was some cheap and nasty stuff I should have said stuff that looked cheap by description but was nasty.

Samba up to and including 3 series has another little fun feature settable locations in the LDAP to look up. The LDAP server could contain multi domains information. So replication from one location to another would be creating backups. Even if the users at each location that could login where different.

Supporting ADS in samba 4 causes the multi domains in a single LDAP to be given up. I am going to miss it.

As you say ch unix was used where appropriate the big important question is where was that. Answer required to know if your network was a Windows NT network or a Samba network with Windows server members in that time frame. To end users of that time-frame both look identical.

This is why I wish Samba 4 with ADS would get out then we could have this fun again.

Who would’ve thought!

“ch I guess you guys out sourced for the solution in 1997 and did not have any unix/bsd staff”

Our company built – among lots of other stuff – servers with their own version of Unix. Yes, we knew Unix – and used it where appropriate.

ch
–Fortunately our admins were more competent than you–
No more competent than the ones I was cleaning up after. I saw some really good ways to ruin a NT.

Most creative was a network card that unless you enabled netbeui it did not work at all. Packaging said 100 Mbs server grade network card. Believing packaging not that wise all the time. Basically cheapest so called server network card was a brick. Yes 1991 to 1999 I have some bad memory of network gear from hell made by different companies on the cheap with nasty big defects. Only lasted 10 years before that the gear was mostly good after that mostly good. You did not have just incompetent IT people you had incompetent hardware makers as well.

–Which implementation of LDAP would you have recommended around 1997 ?–
I will take this as Jan 1997 and before.

Slapd was very popular with us at the time doing BSD bases mostly. Linux servers for me start around 1998. Even that I had been messing with Linux from 1994. No licensing cost either.

So the two dominate would have been NDS and Slapd in 1997. Mostly because NDS for migration so all clients get reconnected to Samba before migrating the user information into Slapd out of NDS. So that zero down time for Netware to Samba was possible.

Even for Linux bases Slapd most likely would have been popular it was the only server for LDAP with out large billing to use yet still a full LDAP 3.2.

Slapd is the base to OpenLDAP today. Slapd was quite mature by 1995. Its also a distant relation of 389 directory server as well.

1998 you list becomes a lot broader.

I guess you missing the 1995 announcement when LDAP become a accepted standard ch. Also missed Slapd release supporting ldap 3.2 in the same year.

It was upto handling the scale you had as 1 domain in 1995 if you trusted it. Of course most businesses would have waited 12 months to let the new techs bugs be fixed.

ch I guess you guys out sourced for the solution in 1997 and did not have any unix/bsd staff so never got offered the more cost effective.

Once again: Your own link disproves you. It’s a recommended limit, not a hard one.

“NT 4 you could by mistake create a Trust circle with two domains named the same at different locations joined with different users creating some very strange permission approvals.”

Fortunately our admins were more competent than you

“LDAP is kinda the key bit here to prevent trust disasters.”

Which implementation of LDAP would you have recommended around 1997 ?

“Thorsten Rahn my knowledge is not Google is really fixing NT up after business had it completely fail.”

Sorry, but your knowledge has been shown to be non-existing a few times too many by now.

Complete Trust Mesh comes in Windows 2000 with ADS does not exist as stable feature in NT 4.

NT 4 has domain to domain trusts that are manually set between domains not a true Trust Mesh. NT 4 you could by mistake create a Trust circle with two domains named the same at different locations joined with different users creating some very strange permission approvals.

Great fun nasty side effects if you pulled it off. Best was seeing NT 4 get lost and give a guest user in one domain equal to admin on another then magically promoting it in its own domain to administrator due to being completely lost in the trust system.

There were a lot of manual made NT 4.0 trust nets that basically exploded with some nicely disaster results for companies.

You are saying features that come in 2000 to explain the extra size. Complete Trust Mesh with sanity checking comes with the ADS. Before that was hold on a hope no one creates a trust circle or use Samba that did not have the bug. LDAP is kinda the key bit here to prevent trust disasters.

NT4 was a stack of independent domains at locations. Not a integrated backbone. That you hope you wired up right with trusts.

NT was much better as an application server. As just a member server NT could do this even with Samba running users.

Thorsten Rahn 1 million+ in a single domain is supported safely by samba. No trust issues either. Samba had trust sanity checking.

So all 400K users could be handled by 1 samba domain without having to setup trust relationships. LDAP replications yes to multi locations and rsync of profiles so allowing users to move between locations and profile follow. This is what I am talking about multi location support. I guess people did not move between offices and have profile follow automatically where you were ch.

Thorsten Rahn and ch I guess where you were that no one made a mistake setting up trusts with NT 4.0 so did not see its complete nasty bad side.

There were reasons why lots of places avoided NT up until 2000. Or tried NT and got badly burnt.

Thorsten Rahn I can tell you back in 1998 I had a few people tell me that yes we could expand SAM past 40 000 objects as well. Let them do it have a script run until 40 000 object are created then have them create one more user and watch the complete thing go down. Never fixed in any of the service packs of NT 4.0.

I guess neither of you were handling networks with insane number of users per domain or where trusts in NT 4.0 went wrong.

Thorsten Rahn my knowledge is not Google is really fixing NT up after business had it completely fail.