Wednesday, February 23, 2005
Gone with the WINS
I've been thinking alot about WINS and browsing these days...
Now, before you start calling Dr. Phil to schedule my first appointment, let me just say that I've actually had *reason* to think about WINS and browsing...
For one, I visit experts-exchange.com semi-regularly, and try to help people out with questions they post there. Inevitably, at least once a day, someone will submit a post which essentially says "blah blah blah, server X can't see server y, blah blah", and I think to myself...browsing again. Inevitably I post a response saying "use WINS". Sometimes I take the time to explain *why* to use WINS, sometimes not. If there's one thing that I've found WINS is good at, it's helping people browse. That said, I'm finding it less and less useful for anything else these days. In fact, I would go so far as to say that except for browsing (which I don't particularly care about anyway), WINS is useless. Yes, I'm calling for it - Death to WINS!
Most of you probably know the story behind all this stuff and how WINS came to be, right? The Coles Notes version is that the essense of what we know as "Windows Networking" was designed and developed circa Windows for Workgroups - and as such, it was made to work in small, peer-to-peer networks that did not span subnets. In that capacity, it worked great - every PC tells every other PC about themselves, and presto, we can all share things and connect to each other's PC in a NetBIOS utopia. Unfortunately, when Microsoft decided to move into the "Enterprise", not-peer-to-peer, more-than-one-subnet arena (think Windows NT), rather than completely scrapping Microsoft Networking and recreating it from scratch, they essentially tried to drag it along and implement workarounds (I'll call them kludges) so that things will still function as they did in that small-network world. Now, some MS developers may bristle at my use of the name kludge, but hey, I'm calling it like I see it. WINS, my friends, is one of those kludges. The fact is that NetBIOS browsing wouldn't work in a multi-subnet environment, so something had to be introduced to make it work, and that something was WINS. WINS allowed NetBIOS name lookups and browsing to span subnets by using client-server and server-to-server calls rather than only broadcasts to do the job. That was great, but of course browsing was still a sick and twisted labyrinth that only two people in the world actually fully understood. OK, I'm being harsh, I'll tackle browsing another day. Oh yes, one of those people (and another impetus for these blogs) is here.
OK, so we have an idea as to why WINS came about, now I'll get to why I think it should (and inevitably will) go away. Again, for the uninitiated, WINS is essentially a database of names and IP addresses. WINS clients query for a name, and the WINS server responds back with the address. The database is dynamically built via registrations that WINS clients sends to the server whenever they start or stop network services. As an aside, acting as a Master Browser or Domain Master Browser is one of those things that gets registered with the WINS server, and *that* is how a Master Browser from one subnet can find a Master Browser from another subnet. The real benefit to WINS in those days was the fact that it was dynamic. We had a different name-lookup service available (DNS), but of course at the time it would have required entering static host records for each and every PC - and either assigning static IP addresses, or setting DHCP reservations for every PC. As you can see, WINS truly was necessary - it was the only name-lookup service that allowed us to use DHCP and didn't require manual upkeep. That was then.
When information started coming out about Windows 2000 and what Active Directory would involve, the deal I heard was that NetBIOS would be gone. This is what I expected - no more NetBIOS names, no more WINS, nada. Obviously I wasn't travelling in the right circles at the time, because as you know, WINS is still alive and well. The problem is, what Windows 2000 did do is to render WINS virtually obsolete by introducing Dynamic DNS.
Now, for the past 5 years we've been living in a world where most organizations are managing and maintaining two parallel and virtually identical network services. Double the admin, double the network traffic, double everything. Show of hands: how many are running both DNS and WINS in your Windows environment? That's what I thought. Do you add static hosts records to both databases? Do you scavenge and backup both databases? Do you configure replication between servers for both services? It's ridiculous really...I'm in a pure Windows 2000/2003 environment - I shouldn't need WINS at all, yet I'm still running it. Why? Browsing. Right now, in a global organization of over 2000 users, we are running a collection of replicating WINS servers for the sole purpose of allowing each and everybody to click and see each and everybody else. Now, in my opinion this is definitely not worth it, but I'll talk about browsing in a later post.
So - in a post NT4, Active-Directory environment, there's really nothing that I need WINS for that DNS can't provide *except* for browsing. The obvious solution to move toward a truly NetBIOS-free world is for Microsoft to implement some equivalent to the functionality of browsing, but using DNS to do it. If you look in 'Entire Network' in 'My Network Places' (it'll always be the Neighbourhood to me...) you'll see "Microsoft Windows Networks" and "Directory". "Microsoft Windows Networks" does not use DNS - this is only populated in a multi-subnet environment by using WINS (or forcing your Master Browsers and adding LMHOSTS entries - but we won't really count that). "Directory" only lets me browse my Domain, and not any other domains that I may or may not trust. Ergo, I still really do need WINS for browsing, but DNS can and is handling everything else for me. Since there are also a number of Linux and Solaris machines on the network, one could even argue that DNS is doing a better job that WINS ever could, just for platform integration reasons alone.
Now - this long-held belief that WINS is useless was very recently shaken by the discovery of this. Unbelievably, Exchange 2000 and Exchange 2003 both *require* WINS for certain functionality to work. That truly blew my mind, and in my opinion can only be considered a flaw in Exchange. Now, we use a competing product instead of Exchange, so this is something I never would have experienced first hand if I didn't come across a link to it somewhere.
I will be participating in an Microsoft Networking MVP-only discussion this week about future networking directions in Longhorn. While I may not be privy to share some of that information here, I'm anxious to here what's in store.
Over the next while we will all start learning more and more about Longhorn and what networking features it will bring. I for one will be watching with a very interested eye any changes to the role that NetBIOS plays in a Mircrosoft network. I look forward to the day when DNS is the one and only name-lookup service we have to deal with, and we truly can say goodbye to WINS!
Now, before you start calling Dr. Phil to schedule my first appointment, let me just say that I've actually had *reason* to think about WINS and browsing...
For one, I visit experts-exchange.com semi-regularly, and try to help people out with questions they post there. Inevitably, at least once a day, someone will submit a post which essentially says "blah blah blah, server X can't see server y, blah blah", and I think to myself...browsing again. Inevitably I post a response saying "use WINS". Sometimes I take the time to explain *why* to use WINS, sometimes not. If there's one thing that I've found WINS is good at, it's helping people browse. That said, I'm finding it less and less useful for anything else these days. In fact, I would go so far as to say that except for browsing (which I don't particularly care about anyway), WINS is useless. Yes, I'm calling for it - Death to WINS!
Most of you probably know the story behind all this stuff and how WINS came to be, right? The Coles Notes version is that the essense of what we know as "Windows Networking" was designed and developed circa Windows for Workgroups - and as such, it was made to work in small, peer-to-peer networks that did not span subnets. In that capacity, it worked great - every PC tells every other PC about themselves, and presto, we can all share things and connect to each other's PC in a NetBIOS utopia. Unfortunately, when Microsoft decided to move into the "Enterprise", not-peer-to-peer, more-than-one-subnet arena (think Windows NT), rather than completely scrapping Microsoft Networking and recreating it from scratch, they essentially tried to drag it along and implement workarounds (I'll call them kludges) so that things will still function as they did in that small-network world. Now, some MS developers may bristle at my use of the name kludge, but hey, I'm calling it like I see it. WINS, my friends, is one of those kludges. The fact is that NetBIOS browsing wouldn't work in a multi-subnet environment, so something had to be introduced to make it work, and that something was WINS. WINS allowed NetBIOS name lookups and browsing to span subnets by using client-server and server-to-server calls rather than only broadcasts to do the job. That was great, but of course browsing was still a sick and twisted labyrinth that only two people in the world actually fully understood. OK, I'm being harsh, I'll tackle browsing another day. Oh yes, one of those people (and another impetus for these blogs) is here.
OK, so we have an idea as to why WINS came about, now I'll get to why I think it should (and inevitably will) go away. Again, for the uninitiated, WINS is essentially a database of names and IP addresses. WINS clients query for a name, and the WINS server responds back with the address. The database is dynamically built via registrations that WINS clients sends to the server whenever they start or stop network services. As an aside, acting as a Master Browser or Domain Master Browser is one of those things that gets registered with the WINS server, and *that* is how a Master Browser from one subnet can find a Master Browser from another subnet. The real benefit to WINS in those days was the fact that it was dynamic. We had a different name-lookup service available (DNS), but of course at the time it would have required entering static host records for each and every PC - and either assigning static IP addresses, or setting DHCP reservations for every PC. As you can see, WINS truly was necessary - it was the only name-lookup service that allowed us to use DHCP and didn't require manual upkeep. That was then.
When information started coming out about Windows 2000 and what Active Directory would involve, the deal I heard was that NetBIOS would be gone. This is what I expected - no more NetBIOS names, no more WINS, nada. Obviously I wasn't travelling in the right circles at the time, because as you know, WINS is still alive and well. The problem is, what Windows 2000 did do is to render WINS virtually obsolete by introducing Dynamic DNS.
Now, for the past 5 years we've been living in a world where most organizations are managing and maintaining two parallel and virtually identical network services. Double the admin, double the network traffic, double everything. Show of hands: how many are running both DNS and WINS in your Windows environment? That's what I thought. Do you add static hosts records to both databases? Do you scavenge and backup both databases? Do you configure replication between servers for both services? It's ridiculous really...I'm in a pure Windows 2000/2003 environment - I shouldn't need WINS at all, yet I'm still running it. Why? Browsing. Right now, in a global organization of over 2000 users, we are running a collection of replicating WINS servers for the sole purpose of allowing each and everybody to click and see each and everybody else. Now, in my opinion this is definitely not worth it, but I'll talk about browsing in a later post.
So - in a post NT4, Active-Directory environment, there's really nothing that I need WINS for that DNS can't provide *except* for browsing. The obvious solution to move toward a truly NetBIOS-free world is for Microsoft to implement some equivalent to the functionality of browsing, but using DNS to do it. If you look in 'Entire Network' in 'My Network Places' (it'll always be the Neighbourhood to me...) you'll see "Microsoft Windows Networks" and "Directory". "Microsoft Windows Networks" does not use DNS - this is only populated in a multi-subnet environment by using WINS (or forcing your Master Browsers and adding LMHOSTS entries - but we won't really count that). "Directory" only lets me browse my Domain, and not any other domains that I may or may not trust. Ergo, I still really do need WINS for browsing, but DNS can and is handling everything else for me. Since there are also a number of Linux and Solaris machines on the network, one could even argue that DNS is doing a better job that WINS ever could, just for platform integration reasons alone.
Now - this long-held belief that WINS is useless was very recently shaken by the discovery of this. Unbelievably, Exchange 2000 and Exchange 2003 both *require* WINS for certain functionality to work. That truly blew my mind, and in my opinion can only be considered a flaw in Exchange. Now, we use a competing product instead of Exchange, so this is something I never would have experienced first hand if I didn't come across a link to it somewhere.
I will be participating in an Microsoft Networking MVP-only discussion this week about future networking directions in Longhorn. While I may not be privy to share some of that information here, I'm anxious to here what's in store.
Over the next while we will all start learning more and more about Longhorn and what networking features it will bring. I for one will be watching with a very interested eye any changes to the role that NetBIOS plays in a Mircrosoft network. I look forward to the day when DNS is the one and only name-lookup service we have to deal with, and we truly can say goodbye to WINS!
