Re: [H] Private IP classes
This is interesting what you say about the Windows file sharing and what not. I always blew off any slowness/stalling when browsing the workgroup to just another crappy bonus of Microsoft Windows. Whenever I would observe this sort of behavior it would happen on one or two workstations but not all of them.. eventually it would seem to straighten up. I will probably do a little experimenting the next time I'm setting up a little Windows based peer-to-peer network. -Tharin O. DHSinclair <[EMAIL PROTECTED]> wrote: I am still reading over your explanation of subnet qualification via the 4th octet of the netmask. Think I'm going to make it into a chart for the wall! This has been a very instructive 2.5 weeks. I am smarter now with your shares, and the shares of j maccraw, but I am still really afraid to try the class C IP addy series. It just does not seem to work here. Perhaps something very subtle on my LAN, and I have not found it in the last 8 years. And, I freely admit it is my fault. The machines are so much "smarter" than I feel sometimes. LOL! Best, Duncan
Re: [H] Private IP classes
inline below. At 09:27 11/01/2007 -0700, you wrote: You can use 10.0.0.1, 10.10.10.1, 10.30.20.1 or any other ip series you want. I just think that you have to use one of the subnet masks mentioned in my previous post that per subnet limits you to 254 or less hosts. Yes, the plan always was to limit hosts to 254 or less. I did realize how big the class A series was. In my case, it just seems to work better than class B or class C. Yes, I remain confused about this a bit. Technically a Class A network would have a default subnet mask of 255.0.0.0. When assigning your router an IP of 10.0.0.1 and a subnet mask of 255.255.255.0 you're still creating a subnet with a maximum of 254 hosts whose ip address would have to be in a range of 10.0.0.1-10.0.0.254. Thank you. This is what I thought and how I proceeded way back in 1999. This is the "pairing" I have always used since moving to w2k/XP. Functionally this should be no different than using a Class C address with the 192.168.x.x ip series. I don't know any reason why one would seem any slower than the other. I'm curious what you mean about being slow or glitchy?? Hmm. OK. Perhaps my IP addy change process is broken. What I did was go to each LAN client/server and use the Control Panel applet to change the IP addy from its' old class A address to its' new class C address. Close the applet and then run cmd/ipconfig /all to check for typos. Go to next client. I did not reboot the clients or server, because w2k/XP does not seem to require this. Perhaps I blew it here.. Once all the clients/servers/router had their new class C IP addy's, I would use the Network Neighborhood function to test the connection between the machines. When addressed in class C, all machines would "stall" opening their shares. Not very long, but there was a noticeable delay before the remote machine displayed its' shares. For the first week of class C, I wrote this delay off to each machine rebuilding its' routing table and/or other internal TCP/IP business. Perhaps my bad! And, I even thought maybe my two switches may be adding latency since they had only been/switched class A addresses since new. Hmm. The only way I know to reset a consumer switch is to PO-PO. I have never been able to talk directly to either of my switches (netgear fs-508 and netgear fs-308) Once I returned everyone to my old class A address scheme (10.0.0.x/255.255.255.0) all of my clients/server/router returned to popping open their shares instantly. The LAN seems to back to its' normal spiffy, speedy, state. I am willing to admit an amount of PEBCAK is involved. By using a Class A address you would just be expanding the theoretical maximum capacity of your private network. A full Class A network subnetted as 255.255.255.0 would create 65536 subnets. I suppose if you had that many $30 routers from Best Buy you could create a private network with over 16 million hosts. H. What a possible power trip Unfortunately I don't have the space or $$$ for this! Interesting idea however. Believe I will pass :) I think in some legacy equipment from... say twenty years ago or more, when this stuff was dreamed up, the logic about class a, b, and c networks would have been hardcoded to some degree. I don't think these kind of designations mean much when creating a private lan. Yes, I though so too. But, as I have seen routers get so much smaller, more intelligent, and feature rich, I feared that I needed to move to the network address scheme the router came default (192.168.0.1). So, I tried. Results were poor, but worked sort of. Initial: Clients were assigned addy's of 192.168.218.x Servers were assigned addy/s of 192.168.219.x Router assigned addy of 192.168.0.1 Printer assigned addy 192.168.219.60 Subnet mask used was 255.255.0.0 Result: Clients could talk to clients, servers talked to servers. Clients and servers could not talk to each other. this made my ESET AV updates on the LAN stop! Hmm. Pilot error! (first email to the List!) Next Try: Clients were assigned addy's of 192.168.218.x Servers were assigned addy/s of 192.168.218.x Router and printer assigned addy's of 192.168.218.x Subnet mask used was 255.255.255.0 Result: Everybody now talks to everybody else, but inter-machine latency is noticeably long. Hmm. This is very odd. But everybody talks to each other, sort of. (2d email to the List!) Current: Clients were assigned addy's of 10.0.0.2-99 Servers were assigned addy/s of 10.0.0.100-199 Router assigned addy of 10.0.0.1 Printer assigned addy of 10.0.0.254 Subnet mask used by all is 255.255.255.0 Result: Everybody now talks to everybody else at normal LAN speed. Inter-machine latency is not an issue at all. I am still reading over your explanation of subnet qualification via the 4th octet of the netmask. Think I'm going to make it into a chart for the wall! This has been a very instructive 2.5
Re: [H] Private IP classes
You can use 10.0.0.1, 10.10.10.1, 10.30.20.1 or any other ip series you want. I just think that you have to use one of the subnet masks mentioned in my previous post that per subnet limits you to 254 or less hosts. Technically a Class A network would have a default subnet mask of 255.0.0.0. When assigning your router an IP of 10.0.0.1 and a subnet mask of 255.255.255.0 you're still creating a subnet with a maximum of 254 hosts whose ip address would have to be in a range of 10.0.0.1-10.0.0.254. Functionally this should be no different than using a Class C address with the 192.168.x.x ip series. I don't know any reason why one would seem any slower than the other. I'm curious what you mean about being slow or glitchy?? By using a Class A address you would just be expanding the theoretical maximum capacity of your private network. A full Class A network subnetted as 255.255.255.0 would create 65536 subnets. I suppose if you had that many $30 routers from Best Buy you could create a private network with over 16 million hosts. I think in some legacy equipment from... say twenty years ago or more, when this stuff was dreamed up, the logic about class a, b, and c networks would have been hardcoded to some degree. I don't think these kind of designations mean much when creating a private lan. -Tharin O. DHSinclair <[EMAIL PROTECTED]> wrote: OK. I understand. I will again attempt to apply a class C network strategy. Even though the LAN runs very much better using class A with a full (25.255.255.0) netmask. Even the router is happy at 10.0.0.1/255.255.255.0.
Re: [H] Private IP classes
inline below. Great share btw.. I'll move ahead and address this share even though I have not readied my return query to the bottom half of your last share.. :) Apologies to the remaining List members if this discussion is causing problems At 11:29 10/31/2007 -0700, you wrote: resending another email that bounced.. did hardwaregroup.com go down yesterday? > > I'm not certain because I've never tested it, but I think on the LAN side you must use a subnet that would be confined to a single Class C network when using a consumer router. Using 255.255.0.0 as a subnet mask would actually be subnetted as a Class B network since only the first two octets would be the network portion. It could be that a router would let you enter this type of subnet on the LAN configuration but would not function as expected. OK. I understand. I will again attempt to apply a class C network strategy. Even though the LAN runs very much better using class A with a full (25.255.255.0) netmask. Even the router is happy at 10.0.0.1/255.255.255.0. Nice choice on the router btw; I own the very same unit. I wanted a router with a gigabit switch and tests on Tom's Hardware showed that it could support a high speed WAN connection as well as many active sessions, thus making it P2P file-sharing friendly. My previous router would spontaneously reboot if I had too many active connections due to Kademlia. It simply wasn't powerful enough to handle the load. Anyway.. back to the subnets.. Superb, another resource for future router questionsHayes, now there are 3 of us!. :) I think the only valid choices for a subnet mask when using these kinds of routers would be one that restricts you to functioning under a single Class C network. (when I say Class C I mean that when looking at an IP of AAA.BBB.CCC.xxx only hosts whose first three octets are matching, can communicate directly) The usable subnet masks are the following 255.255.255.0 (1 network, 254 hosts) 255.255.255.128 (2 networks, 126 hosts each) 255.255.255.192 (4 networks, 62 hosts each) 255.255.255.224 (8 networks, 30 hosts each) 255.255.255.240 (16 networks, 14 hosts each) 255.255.255.248 (32 networks, 6 hosts each, used with 5 ip accounts) 255.255.255.252 (64 networks, 2 hosts each, what most ISPs assign) OMG. I did miss this business 2 years ago. So, this is how true "subnets" - actual seperate channels for groups of PCs get done. Amazing, truly amazing. I do not fully grasp, but now know which chapters of my book to re-read. :) Now, I just finished trying to use a 192.168.218.x IP series. Maybe boneheaded, but I used the value of 218 in the 3rd octet just because my home address is "218." It seemed like a valid way to differentiate "my" LAN from all of the plethora of 192.168.?.? LANs on planet Earth. The fun started from this decision. It took, but was slow and glitchy. When I make a major LAN change to my LAN topography (like an IP address series choice), should I be using some tool or "reset process" for the winsock when I make these changes (class A to class C)? I never did. And still have not. Even though I am back to almost where I started with some minor address changes. If you were using a Class C address of 192.168.1.x and a subnet mask of 255.255.255.192 you would essentially be chopping up 192.168.1.x into 4 subnets. There would be a total of 64 ip addresses in each subnet but since the first and last address of any network are reserved, there are only 62 usable addresses for your hosts. The same logic can be seen in the other subnet masks. I know I'm probably repeating/rephrasing some of what I already said but I think more examples help me when trying to understand something. Do not apologize, please! Your examples help draw the mental pictures I need to grasp these really subtle things. I was working for Xerox back when this "ethernet" thingy came to be. Thought I had a good grasp. Perhaps not. It just seems to have gotten so much more complicated - for obvious reasons! I do appreciate the level of detail. Best, Duncan -Tharin O. DHSinclair <[EMAIL PROTECTED]> wrote: Tharin, Thank you for the reply. The smoke clears. I want to read your reply a few more times.
Re: [H] Private IP classes
resending another email that bounced.. did hardwaregroup.com go down yesterday? > > I'm not certain because I've never tested it, but I think on the LAN side you must use a subnet that would be confined to a single Class C network when using a consumer router. Using 255.255.0.0 as a subnet mask would actually be subnetted as a Class B network since only the first two octets would be the network portion. It could be that a router would let you enter this type of subnet on the LAN configuration but would not function as expected. Nice choice on the router btw; I own the very same unit. I wanted a router with a gigabit switch and tests on Tom's Hardware showed that it could support a high speed WAN connection as well as many active sessions, thus making it P2P file-sharing friendly. My previous router would spontaneously reboot if I had too many active connections due to Kademlia. It simply wasn't powerful enough to handle the load. Anyway.. back to the subnets.. I think the only valid choices for a subnet mask when using these kinds of routers would be one that restricts you to functioning under a single Class C network. (when I say Class C I mean that when looking at an IP of AAA.BBB.CCC.xxx only hosts whose first three octets are matching, can communicate directly) The usable subnet masks are the following 255.255.255.0 (1 network, 254 hosts) 255.255.255.128 (2 networks, 126 hosts each) 255.255.255.192 (4 networks, 62 hosts each) 255.255.255.224 (8 networks, 30 hosts each) 255.255.255.240 (16 networks, 14 hosts each) 255.255.255.248 (32 networks, 6 hosts each, used with 5 ip accounts) 255.255.255.252 (64 networks, 2 hosts each, what most ISPs assign) If you were using a Class C address of 192.168.1.x and a subnet mask of 255.255.255.192 you would essentially be chopping up 192.168.1.x into 4 subnets. There would be a total of 64 ip addresses in each subnet but since the first and last address of any network are reserved, there are only 62 usable addresses for your hosts. The same logic can be seen in the other subnet masks. I know I'm probably repeating/rephrasing some of what I already said but I think more examples help me when trying to understand something. -Tharin O. DHSinclair <[EMAIL PROTECTED]> wrote: Tharin, Thank you for the reply. The smoke clears. I want to read your reply a few more times.
Re: [H] Private IP classes
Tharin, Thank you for the reply. The smoke clears. I want to read your reply a few more times. Yes, I did mis-speak, should have used "octets" instead of "quartiles." Got it! The first paragraph makes perfect sense. Got it. Second paragraph makes sense also, but I was advised to use the 255.255.0.0 subnet mask. I did try it. It did not work well. I now use a netmask of 255.255.255.0. This works perfectly now even though I have chosen to use the class A series. Paragraph 3, about the router is still confusing. If I gave my router/gateway an IP addy of 192.168.1.1, would I not also have to use a netmask of 255.255.0.0 for the 3rd and 4th octets to have any meaning/control? Do not know how to answer Paragraph 4. Who/What would I "share" my xdsl link with? I do use a router, a DLink DGL-4300. The recc came from Hayes. I chose it mainly for its' strong SPI firewall. It does offer WIFY, but I have that disabled. I am a hard-wire guy. Yes, you are making very much sense. It has just been so many years since I had broadband, I'm afraid I forgot much of the edge I had back in 1999-2000 from previous List training. I will come back later regarding the balance of your very good share. ATM, the LAN, router, and clients all seem to be happy after late last night's re-address party! More later after study. Thank you so much. Best, Duncan At 10:38 10/30/2007 -0700, you wrote: Simple answer is that a Private Class C network with a 255.255.255.0 subnet mask is fine for you. This would create a local network allowing up to 254 hosts (machines, computers, whatever). If my math is right, 254 is way bigger than than 7 :) When using this subnet you must make sure the first three quartiles are the exact same. All hosts need to have matching IPs except for the fourth set. If you want to make use of the entire block of 192.168.x.x then use 255.255.0.0. (But why would you need a lan of 65 thousand addresses?) Customarily a router/gateway is assigned an IP like 192.168.1.1 or 192.168.1.254, since 1 and 254 are the first and last IP address that can be used in the last quartile. Your computers, slingbox, nintendo wii, voip phone, and like can then be assigned addresses ranging from 1-253 or 2-254 depending on which one you choose for your gateway. Are you going to share a DSL or Cable internet connection? Do you already have a router or were you turning a PC into a server/gateway/router? ~~ More drawn own discussion below ~~ Since computers use binary the proper term, for what you refer to as a quartile, is actually an octet. Binary is ugly for people to read and remember so we write IPs in a decimal form 'aaa.bbb.ccc.ddd' Anyway, the local network is defined by the subnet and ip that you assign. If you use 255.255.255.0 as a subnet mask with a Class C network, then essentially you are dividing up 192.168.x.x into 256 local networks that can each have 254 hosts. A computer/machine/host/whatever will only be able to talk to others that are using an IP that match the first three octets 192.168.1.x, 192.168.2.x, 192.168.3.x, etc. ('x' can be a value from 1 to 254, 0 and 255 are reserved). A computer on 192.168.2.x can't talk to a computer on 192.168.3.x. Data bound for an address outside the scope of your local network, such as the internet, would have the packets sent to your router/gateway. (The gateway should be your router. A consumer router is actually several things its a gateway, a router, a switch, a firewall, and often times a wireless access point.) If you were to use a different subnet like 255.255.255.240 and it would create a local network with 16 addresses. This would means you could have 16 subnets on a single Class C block. 192.168.1.0-16 (LAN 1), 192.168.1.17-32 (LAN 2), 192.168.1.33-48 (LAN 3), etc. The entire 192.168.x block would be divided into 4000+ seperate networks. A subnet of 255.255.0.0 would turn the whole 192.168 block into your local network. Any machine with an ip in the range of 192.168.0-255.x could talk to each other. I hope I'm making some sense... -Tharin O. DHSinclair <[EMAIL PROTECTED]> wrote: Tharin, Can I ask for some more expansion on the below send? This may be the critical link! At 16:02 10/29/2007 -0700, you wrote: >snip >The subnet mask will determine the range of IP addresses that will be in >your local network. Your local network being the computers/devices you >have direct access to send data without needing to be handled through a router. > >snip >How did you need to apply this knowledge?? Are you configuring a router or >small network in your home?? I was trying to have 2 subnets. Perhaps I did it wrong/badly. I have given up! Why is the "...needing to be handled through a router" part of this equation? This may be my confusion Best, Duncan snip
Re: [H] Private IP classes
Simple answer is that a Private Class C network with a 255.255.255.0 subnet mask is fine for you. This would create a local network allowing up to 254 hosts (machines, computers, whatever). If my math is right, 254 is way bigger than than 7 :) When using this subnet you must make sure the first three quartiles are the exact same. All hosts need to have matching IPs except for the fourth set. If you want to make use of the entire block of 192.168.x.x then use 255.255.0.0. (But why would you need a lan of 65 thousand addresses?) Customarily a router/gateway is assigned an IP like 192.168.1.1 or 192.168.1.254, since 1 and 254 are the first and last IP address that can be used in the last quartile. Your computers, slingbox, nintendo wii, voip phone, and like can then be assigned addresses ranging from 1-253 or 2-254 depending on which one you choose for your gateway. Are you going to share a DSL or Cable internet connection? Do you already have a router or were you turning a PC into a server/gateway/router? ~~ More drawn own discussion below ~~ Since computers use binary the proper term, for what you refer to as a quartile, is actually an octet. Binary is ugly for people to read and remember so we write IPs in a decimal form 'aaa.bbb.ccc.ddd' Anyway, the local network is defined by the subnet and ip that you assign. If you use 255.255.255.0 as a subnet mask with a Class C network, then essentially you are dividing up 192.168.x.x into 256 local networks that can each have 254 hosts. A computer/machine/host/whatever will only be able to talk to others that are using an IP that match the first three octets 192.168.1.x, 192.168.2.x, 192.168.3.x, etc. ('x' can be a value from 1 to 254, 0 and 255 are reserved). A computer on 192.168.2.x can't talk to a computer on 192.168.3.x. Data bound for an address outside the scope of your local network, such as the internet, would have the packets sent to your router/gateway. (The gateway should be your router. A consumer router is actually several things its a gateway, a router, a switch, a firewall, and often times a wireless access point.) If you were to use a different subnet like 255.255.255.240 and it would create a local network with 16 addresses. This would means you could have 16 subnets on a single Class C block. 192.168.1.0-16 (LAN 1), 192.168.1.17-32 (LAN 2), 192.168.1.33-48 (LAN 3), etc. The entire 192.168.x block would be divided into 4000+ seperate networks. A subnet of 255.255.0.0 would turn the whole 192.168 block into your local network. Any machine with an ip in the range of 192.168.0-255.x could talk to each other. I hope I'm making some sense... -Tharin O. DHSinclair <[EMAIL PROTECTED]> wrote: Tharin, Can I ask for some more expansion on the below send? This may be the critical link! At 16:02 10/29/2007 -0700, you wrote: >snip >The subnet mask will determine the range of IP addresses that will be in >your local network. Your local network being the computers/devices you >have direct access to send data without needing to be handled through a router. > >snip >How did you need to apply this knowledge?? Are you configuring a router or >small network in your home?? I was trying to have 2 subnets. Perhaps I did it wrong/badly. I have given up! Why is the "...needing to be handled through a router" part of this equation? This may be my confusion Best, Duncan snip
Re: [H] Private IP classes
Thank you j maccraw, That became the proof at about 0130 this morning. LOL! I now have a cheat sheet that says exactly what you shared. I'll update it with the correct word "octet." All is back up and running much quicker with proper addy's and netmasks. Would I be correct that when using the netmask of 255.255.255.0, the network id becomes the 1st-3rd octets, and the machine id is the 4th octet? Best, Duncan At 01:41 10/30/2007 -0700, j maccraw wrote: Netmasks determine what bits of the 32-bit address is network id vs. machine id. Whatever you choose for the 1st 3 octets with a netmask of 255.255.255.0 IS important only in that all machines must use the same 1st 3 octets. 192.168.0.1 - 192.168.0.254. If you switch to a 255.255.0.0 netmask, then only the 1st 2 octets are important. 192.168.0.1 - 192.168.255.254. > DHSinclair <[EMAIL PROTECTED]> wrote: OK, I now use what I read is the Class C private IP address series. > I use 192.168.2xx.x. I am told that my sub-net mask should > be 255.255.255.0. > If this is true, then should not my chosen 3rd quartile of "2xx" really > be a value 0? > Or, does it really mean that my chosen 3rd quartile is somehow > ignored? Yes, I do recall discussion about this last year, but back > then I was using the Class A private IP series. :) > Thank you for any/all de-mystification. > Best, > Duncan > > > __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
Re: [H] Private IP classes
Netmasks determine what bits of the 32-bit address is network id vs. machine id. Whatever you choose for the 1st 3 octets with a netmask of 255.255.255.0 IS important only in that all machines must use the same 1st 3 octets. 192.168.0.1 - 192.168.0.254. If you switch to a 255.255.0.0 netmask, then only the 1st 2 octets are important. 192.168.0.1 - 192.168.255.254. > DHSinclair <[EMAIL PROTECTED]> wrote: OK, I now use what I read is the Class C private IP address series. > I use 192.168.2xx.x. I am told that my sub-net mask should > be 255.255.255.0. > If this is true, then should not my chosen 3rd quartile of "2xx" really > be a value 0? > Or, does it really mean that my chosen 3rd quartile is somehow > ignored? Yes, I do recall discussion about this last year, but back > then I was using the Class A private IP series. :) > Thank you for any/all de-mystification. > Best, > Duncan > > > __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
Re: [H] Private IP classes
Tharin, Can I ask for some more expansion on the below send? This may be the critical link! At 16:02 10/29/2007 -0700, you wrote: snip The subnet mask will determine the range of IP addresses that will be in your local network. Your local network being the computers/devices you have direct access to send data without needing to be handled through a router. snip How did you need to apply this knowledge?? Are you configuring a router or small network in your home?? I was trying to have 2 subnets. Perhaps I did it wrong/badly. I have given up! Why is the "...needing to be handled through a router" part of this equation? This may be my confusion Best, Duncan snip
Re: [H] Private IP classes
Tharin, I am trying to configure both a router and my personal network of 6 (7) devices. 7 if I count the router; which I do. Does that combination have a bearing on this? After 7 years, I was trying to stretch/play a little bit because this is my 2d go-around with high-speed. Perhaps I went too far. I tried to use the 3rd quartile/block of the IP address to define a "server" and a "client PC." It did not work at all. "Servers" could talk to/see "servers but not clients. "Clients" could talkto/see "clients" but not "servers". Yes, I was trying to use the 3rd quartile/block of the IP address to differentiate "servers" and "clients." Well, no matter. OK, I may now back up and re-implement my old plan of using a Class A network IP address scheme, but use a sub-net mask of 255.255.255.0. This scheme served me well for the last 7 years. You did answer my question and confirmed I did have a grasp about the "sub-net mask" control. I chose the Class C scheme due to comment here on the list. It just does not work with my chosen IP address series, but now I know that my chosen IP address series may be 'out of bounds.' I will play some more tonight. I will get it to work. More play in process.. Thank you. Best/Most Crisp answer so far... :) Best, Duncan At 16:02 10/29/2007 -0700, you wrote: There is something like 64 thousand ip addresses in the private 192.168.x.x range. (192.168.0.0 - 192.168.255.255) The subnet mask will determine the range of IP addresses that will be in your local network. Your local network being the computers/devices you have direct access to send data without needing to be handled through a router. The common and default subnet mask that is used in a SOHO network and on a Private C class block is 255.255.255.0. This will allow for a total of 254 hosts. This is the default setting for most consumer routers from Linksys, D-link, Netgear, etc. If computerA had an ip address of 192.168.0.12 and a subnet mask of 255.255.255.0 it would be able to communicate with the network range of 192.168.0.1-192.168.0.254. If computerB had an IP of 192.168.9.5 and the same subnet the valid network range would be 192.168.9.1-192.168.9.254. A ping from computerA would not reach computerB or vice versa because they are on two different networks. Hopefully this crude example shows how the subnet mask determines the ip range; especially, how that particular subnet mask would restrict your ip range to whichever block you choose in the third set of numbers. How did you need to apply this knowledge?? Are you configuring a router or small network in your home?? DHSinclair <[EMAIL PROTECTED]> wrote: OK, I now use what I read is the Class C private IP address series. I use 192.168.2xx.x. I am told that my sub-net mask should be 255.255.255.0. If this is true, then should not my chosen 3rd quartile of "2xx" really be a value 0? Or, does it really mean that my chosen 3rd quartile is somehow ignored? Yes, I do recall discussion about this last year, but back then I was using the Class A private IP series. :) Thank you for any/all de-mystification. Best, Duncan
Re: [H] Private IP classes
There is something like 64 thousand ip addresses in the private 192.168.x.x range. (192.168.0.0 - 192.168.255.255) The subnet mask will determine the range of IP addresses that will be in your local network. Your local network being the computers/devices you have direct access to send data without needing to be handled through a router. The common and default subnet mask that is used in a SOHO network and on a Private C class block is 255.255.255.0. This will allow for a total of 254 hosts. This is the default setting for most consumer routers from Linksys, D-link, Netgear, etc. If computerA had an ip address of 192.168.0.12 and a subnet mask of 255.255.255.0 it would be able to communicate with the network range of 192.168.0.1-192.168.0.254. If computerB had an IP of 192.168.9.5 and the same subnet the valid network range would be 192.168.9.1-192.168.9.254. A ping from computerA would not reach computerB or vice versa because they are on two different networks. Hopefully this crude example shows how the subnet mask determines the ip range; especially, how that particular subnet mask would restrict your ip range to whichever block you choose in the third set of numbers. How did you need to apply this knowledge?? Are you configuring a router or small network in your home?? DHSinclair <[EMAIL PROTECTED]> wrote: OK, I now use what I read is the Class C private IP address series. I use 192.168.2xx.x. I am told that my sub-net mask should be 255.255.255.0. If this is true, then should not my chosen 3rd quartile of "2xx" really be a value 0? Or, does it really mean that my chosen 3rd quartile is somehow ignored? Yes, I do recall discussion about this last year, but back then I was using the Class A private IP series. :) Thank you for any/all de-mystification. Best, Duncan