Networking Questions - IBM AS400

This is a discussion on Networking Questions - IBM AS400 ; Much to my regret, I am new to tcp/ip LANS. I need to have a basic understanding of a couple of things. (1) Address setting... Static or dynamic in a private network with fewer than 30 users. Do I really ...

+ Reply to Thread
Results 1 to 17 of 17

Thread: Networking Questions

  1. Networking Questions

    Much to my regret, I am new to tcp/ip LANS.

    I need to have a basic understanding of a couple of things.

    (1) Address setting... Static or dynamic in a private network
    with fewer than 30 users. Do I really need DNS for a simple
    network ? (IBM manuals are worthless in design help)

    The book wants you to put all the pc's in the HOST file but
    again, Access seems to work fine even though the pc is not
    in the HOST file.

    (2) Using a static address as the Iseries Access manual suggests
    works OK but I found that the customer's DSL line gets all their
    address's from a DNS server. When you fire up the internet connect,
    it destroys the static address I assigned.

    How do I handle that problem ???

    (3) We are having a internet fiber link installed for this customer and
    I'm sure this will add even more problems to this entire address
    problem.

    (4) Do I need a "HOST" file on every single PC as the Access manuals
    suggests ? Why ?

    Thanks for any suggestions....

    who has no desire to be a network admin...



  2. Re: Networking Questions

    Hi pat

    Personally I tend to set all of the IP addresses as static, but if you
    must use DHCP, then reserve a range for your 400 and any network
    attached laser printers.

    Perhaps, 192.168.1.1-100 available for DHCP

    192.168.1.101-190 for printers

    192.168.1.200 for AS400

    192.168.1.199 for your router

    You will also need to set up a *DFTROUTE on the 400 of 192.168.1.199
    for the router to handle remote access.

    HTH.

    Cheers
    Neil


  3. Re: Networking Questions


    Just to clarify, your printers, AS400 and router would allways be
    static.

    Cheers
    Neil


  4. Re: Networking Questions

    That's what I thought but what do I do about the DSL line
    and the Windows "wizard" that changes the static address
    for the internet connection ???

    Apparently the DHCP or DNS of the ISP supplies all the
    address's which wipes out my static address....

    There has to be a way around that...

    Newbie_Neil wrote:

    > Just to clarify, your printers, AS400 and router would allways be
    > static.
    >
    > Cheers
    > Neil
    >


  5. Re: Networking Questions

    I am with you right up to the *dftroute bit...

    Why are you using that particular address as
    the "remote access" ??? What exactly does default route
    do ???

    I should also explained that the Iseries is on a LAN
    along with all these other pc's. There is no plans to
    have the Iseries exposed directly to the web. The users
    will have web(selected) access and email that will be
    provided by the ISP. No email on the Iseries at all.


    Newbie_Neil wrote:

    > Hi pat
    >
    > Personally I tend to set all of the IP addresses as static, but if you
    > must use DHCP, then reserve a range for your 400 and any network
    > attached laser printers.
    >
    > Perhaps, 192.168.1.1-100 available for DHCP
    >
    > 192.168.1.101-190 for printers
    >
    > 192.168.1.200 for AS400
    >
    > 192.168.1.199 for your router
    >
    > You will also need to set up a *DFTROUTE on the 400 of 192.168.1.199
    > for the router to handle remote access.
    >
    > HTH.
    >
    > Cheers
    > Neil
    >


  6. Re: Networking Questions

    > I am with you right up to the *dftroute bit...
    >
    > Why are you using that particular address as
    > the "remote access" ??? What exactly does default route
    > do ???


    Someone'll correct me if I've got this wrong... but the default route
    allows the 400 to access IP addresses outside of your current
    "segment" of the network.

    For example, if your 400's IP was 192.168.1.200 and it tried to access
    a device, say at 192.168.2.50, it wouldn't (typically) be able to find
    it - the 400 is in 192.168.1 segment and the other device in
    192.168.2. So the 400 sends the request to the default route - your
    router. If the router cannot find it in any network segments it knows
    of, it'll forward it on to its default route (possibly another router
    or even to the internet - depends on the router setup).

    Sorry if this is an over simplification, but I hope you get the
    idea :-)

    However, it probably wont matter to you as you've only got a small
    network with 30 or so devices,

    HTH.


  7. Re: Networking Questions

    I think this is part of the problem I'm having...

    I set the static address for Access but when we need
    to go to the web, the isp DNS creates a new address
    for the terminal.

    I assume I can set it up so that we can directly to
    the ISP but if the ISP is providing the address, that
    wrecks my static address.

    The *dftroute explanation does make sense...

    xyzzy wrote:

    >>I am with you right up to the *dftroute bit...
    >>
    >>Why are you using that particular address as
    >>the "remote access" ??? What exactly does default route
    >>do ???

    >
    >
    > Someone'll correct me if I've got this wrong... but the default route
    > allows the 400 to access IP addresses outside of your current
    > "segment" of the network.
    >
    > For example, if your 400's IP was 192.168.1.200 and it tried to access
    > a device, say at 192.168.2.50, it wouldn't (typically) be able to find
    > it - the 400 is in 192.168.1 segment and the other device in
    > 192.168.2. So the 400 sends the request to the default route - your
    > router. If the router cannot find it in any network segments it knows
    > of, it'll forward it on to its default route (possibly another router
    > or even to the internet - depends on the router setup).
    >
    > Sorry if this is an over simplification, but I hope you get the
    > idea :-)
    >
    > However, it probably wont matter to you as you've only got a small
    > network with 30 or so devices,
    >
    > HTH.
    >


  8. Re: Networking Questions

    > I set the static address for Access but when we need
    > to go to the web, the isp DNS creates a new address
    > for the terminal.


    I don't quite follow what is happening here. Open up a dos prompt
    (under Win2000 or XP, Start | Run | cmd) whilst you have a static
    address, then type IPCONFIG and note down the ip address, subnet mask
    and default gateway. Next, go to the web. Then at the dos prompt, run
    IPCONFIG again. Have any of the values changed? You could also try
    pinging the 400 whilst you have a static address and after accessing
    the web - from the dos prompt enter: PING AS400_IP_ADDRESS. What
    response do you get from ping on both occasions?


  9. Re: Networking Questions

    Apparently the DSL connection uses DNS to get it's address.

    I did exactly what you said...

    I enter the static address 192.168.1.1 ---> new PC
    255.255.255.0

    No DNS address.

    I can access the Iseries with Access fine.
    I close Iseries Access.

    I start the internet connection on the PC(XP) and it
    fires up a connection.

    I go back and check IPCONFIG and my address is now:

    192.168.17.64
    255.255.255.0

    It also shows a DNS address at this point.

    gateway address is 192.168.17.254

    It's like magic...(normal for MS products)



    xyzzy wrote:

    >>I set the static address for Access but when we need
    >>to go to the web, the isp DNS creates a new address
    >>for the terminal.

    >
    >
    > I don't quite follow what is happening here. Open up a dos prompt
    > (under Win2000 or XP, Start | Run | cmd) whilst you have a static
    > address, then type IPCONFIG and note down the ip address, subnet mask
    > and default gateway. Next, go to the web. Then at the dos prompt, run
    > IPCONFIG again. Have any of the values changed? You could also try
    > pinging the 400 whilst you have a static address and after accessing
    > the web - from the dos prompt enter: PING AS400_IP_ADDRESS. What
    > response do you get from ping on both occasions?
    >


  10. Re: Networking Questions

    > I did exactly what you said...
    >
    > I enter the static address 192.168.1.1 ---> new PC
    > 255.255.255.0
    >
    > No DNS address.
    >
    > I can access the Iseries with Access fine.
    > I close Iseries Access.
    >
    > I start the internet connection on the PC(XP) and it
    > fires up a connection.
    >
    > I go back and check IPCONFIG and my address is now:
    >
    > 192.168.17.64
    > 255.255.255.0
    >
    > It also shows a DNS address at this point.
    >
    > gateway address is 192.168.17.254


    Hmmmmm..... how are you setting the static IP address? I'm unsure as
    to why it is changing once you have accessed the web. If I set a
    static IP address on my PC it does not change even after accessing the
    web. Try running the command IPCONFIG /ALL before and after accessing
    the web. It gives more info. Is there a DHCP server listed there? I
    think the DNS issue may be a red herring - DNS is used to translate a
    name into an address. i.e it translates something like www.google.co.uk
    into the IP address 216.239.59.103.



  11. Re: Networking Questions

    Pat:

    Looks like you'll be learning from experience like many of us did.

    Comments added..

    Pat Barber wrote:

    > Apparently the DSL connection uses DNS to get it's address.


    Not exactly. DHCP (Dynamic Host Configuration Protocol) is what supplies
    the address. The DNS address is sent as a secondary element, mostly
    because there's no point for nearly all internet connections without
    DNS. Its address to the PC when it requests an address for the internet
    connection. The PC asks for an address by sending a DHCP request out the
    route to the DSL device which is either a modem or a router.

    > I did exactly what you said...
    >
    > I enter the static address 192.168.1.1 ---> new PC
    > 255.255.255.0
    >
    > No DNS address.


    No need for DNS until host names get involved and those hosts are on a
    different network segment. Hosts on the same segment can have their
    addresses resolved by a lookup in a local "hosts" table. DNS is mostly
    just a very glorified hosts table that includes addresses in all network
    address ranges. (Well, most address ranges. Some are reserved for local
    use only and a few other special purposes.)

    Note that some address ranges are not allowed to communicate across the
    internet (at least, not without special handling).

    > I can access the Iseries with Access fine.
    > I close Iseries Access.
    >
    > I start the internet connection on the PC(XP) and it
    > fires up a connection.
    >
    > I go back and check IPCONFIG and my address is now:
    >
    > 192.168.17.64
    > 255.255.255.0
    >
    > It also shows a DNS address at this point.
    >
    > gateway address is 192.168.17.254
    >
    > It's like magic...(normal for MS products)


    It's essentially unrelated to MS. That's how TCP/IP and related
    protocols work.

    What I do at home is put a router for my house network between my DSL
    router and my network. My PCs and other devices don't communicate with
    the DSL router; only my personal router does. My router requests a
    dynamic address for itself. My PCs keep their addresses.

    Also, my router does NAT; so it handles 'N'etwork 'A'ddress
    'T'ranslation, translating my PC addresses to whatever addresses are
    handed out through the DSL router. My PCs don't even know that the
    addresses seen by the DSL are unrelated to the addresses known to them.

    This PC I'm working on wants to talk to the internet, so it sends a
    request to an address that doesn't exist on my network out the wire. My
    local router is the default route, so it gets the request because
    there's no place on my network with that remote address. Because it's a
    router, it routes the request out through its uplink to the DSL router.
    It translates the PC address into the NAT address that it associated
    with my PC, so any replies that come back will be tagged with the NAT
    address. Replies come back tagged with the NAT address, and the router
    knows it should translate that back to the original PC address before
    putting the reply into my local network.

    Technically, I don't need to know anything about what the DSL router
    does. I'm pretty sure that it's also doing its own NATting, so this PC
    has its address translated twice. But it's really irrelevant to me. I
    leave that up to the routers.

    In any case, my local router is what keeps the internet/DSL connection
    alive. My PCs never "connect" to DSL. The DSL is always "on". From the
    PCs' viewpoints (or my AS/400's), that's simply another network segment.

    As far as DNS goes, my ISP's DNS address gets inserted into my PCs right
    after one gets powered on and starts talking to the network. I don't
    really know what the real DNS address is; it's most likely NATted by the
    DSL anyway. My PCs are actually using DHCP, but the requests are handled
    by my local router and assigned within the local address range that I
    chose. So far, the address for my ISP's DNS has never changed, so AFAIK
    I could specify it as a static address.

    Because I use DHCP and the dynamic addresses are assigned by my local
    router, the DNS address is also assigned by my router. However, it got
    the address from DSL when it was powered on because I configured the
    router to get its external address by DHCP from DSL.

    I turn the router on; it asks the DSL router for an address to use for
    itself and an address for DNS. That DNS address gets copied to any PC
    when the PC powers up.

    DNS and host tables have one particular use -- to allow you to use names
    instead of IP addresses. If you never refer to a remote system by name,
    you never need DNS nor host tables. A whole bunch of names get different
    addresses at regular (or irregular) intervals. Because addresses change,
    names are handy. DNS automates the association of name with address.

    For a local network, addresses can work fine. You don't really need host
    tables on your PCs; they exist for convenience. I don't have a local
    DNS, but I do have my AS/400 at a fixed (static) local address. I
    usually store that name/address pair in a host table on my PCs, but
    that's generally the only address in the PC host table.

    Well, that's not true. I actually have a bunch of addresses in this PC's
    host table. Here are some of them:

    127.0.0.1 a.tribalfusion.com
    127.0.0.1 ad.doubleclick.net
    127.0.0.1 servedby.advertising.com
    127.0.0.1 ad.targetingmarketplace.com
    127.0.0.1 media.fastclick.net
    127.0.0.1 adserver.00web.com
    127.0.0.1 anad.tacoda.net
    127.0.0.1 adopt.euroclick.com
    127.0.0.1 c.casalemedia.com

    You might recognize one or two.

    I collect them. They're representative of host names for many junk ads
    that you see on lots of sites. When I find one that irritates me, I add
    the complete host name to my hosts table with the localhost address
    127.0.0.1. When a web page refers to that host, the web browser looks at
    the hosts table first. Since it finds a match, it sends the request to
    the localhost address instead of whatever the real address is. The
    localhost address has no clue what's being requested, so the request
    gets rejected fairly quickly. Whatever image would have been sent never
    really gets requested. The web page just has a blank spot there. The
    pages show up a lot faster because there's less junk on it, and no
    images have to take my bandwidth and no image gets copied into my cache.

    Sorry, for getting carried away with this response. The whole point is
    that your PCs shouldn't have to "open" a connection through DSL except
    when they're first turned on. The connection can simply "be there". It
    can simply be available at all times.

    In my case, it's there because I kind of off-loaded that work to a
    fairly cheap piece of hardware -- a cheap Linksys router.

    It's actually more complicated than that because (1) I have two local
    network routers, (2) the other one is a bit more expensive than the
    Linksys, (3) both of them are also wireless routers, (4) this PC has two
    ethernet ports, one of which actually connects direct to the DSL router
    and the other to my local router, (5) I have a KVM switch right here for
    a second PC that has an ethernet port to my local router and also
    connects via dialup to Earthlink, Netscape.com or Compuserve, (6) my
    home network includes segments that are switched through the power lines
    throughout my house, (7) four switches at different parts of the house
    handle some small collections of devices...

    That is, once you get started experimenting, things just start sprouting
    up all over. As I started this post out, it's all just experimenting
    until it starts making sense. I'm hoping it'll make sense to me soon.

    --
    Tom Liotta
    http://zap.to/tl400

  12. Re: Networking Questions

    I know your name from the old News/400 magazine.

    I appreciate your rather generous explanation of
    what you have done.

    This entire tcp/ip networking thing has been a real
    eye-opener for me. D.I.Y. networking is not my thing
    but I will figure it out in time.

    I much prefer the system side of things and I might also
    say that SNA/SDLC makes tcp/ip look very crude in comparison.

    Waaaay too many layers of obscurity, not to mention total
    confusion.

    When I think back to the S/34 and just setting an address and
    the system handled it makes this look pretty sad.

    I agree that TCP/IP is the new wave but what a giant leap
    backward.

    Thomas wrote:

    > Pat:
    >
    > Looks like you'll be learning from experience like many of us did.
    >
    > Comments added..
    >
    > Pat Barber wrote:
    >
    >> Apparently the DSL connection uses DNS to get it's address.

    >
    >
    > Not exactly. DHCP (Dynamic Host Configuration Protocol) is what supplies
    > the address. The DNS address is sent as a secondary element, mostly
    > because there's no point for nearly all internet connections without
    > DNS. Its address to the PC when it requests an address for the internet
    > connection. The PC asks for an address by sending a DHCP request out the
    > route to the DSL device which is either a modem or a router.
    >
    >> I did exactly what you said...
    >>
    >> I enter the static address 192.168.1.1 ---> new PC
    >> 255.255.255.0
    >>
    >> No DNS address.

    >
    >
    > No need for DNS until host names get involved and those hosts are on a
    > different network segment. Hosts on the same segment can have their
    > addresses resolved by a lookup in a local "hosts" table. DNS is mostly
    > just a very glorified hosts table that includes addresses in all network
    > address ranges. (Well, most address ranges. Some are reserved for local
    > use only and a few other special purposes.)
    >
    > Note that some address ranges are not allowed to communicate across the
    > internet (at least, not without special handling).
    >
    >> I can access the Iseries with Access fine.
    >> I close Iseries Access.
    >>
    >> I start the internet connection on the PC(XP) and it
    >> fires up a connection.
    >>
    >> I go back and check IPCONFIG and my address is now:
    >>
    >> 192.168.17.64
    >> 255.255.255.0
    >>
    >> It also shows a DNS address at this point.
    >>
    >> gateway address is 192.168.17.254
    >>
    >> It's like magic...(normal for MS products)

    >
    >
    > It's essentially unrelated to MS. That's how TCP/IP and related
    > protocols work.
    >
    > What I do at home is put a router for my house network between my DSL
    > router and my network. My PCs and other devices don't communicate with
    > the DSL router; only my personal router does. My router requests a
    > dynamic address for itself. My PCs keep their addresses.
    >
    > Also, my router does NAT; so it handles 'N'etwork 'A'ddress
    > 'T'ranslation, translating my PC addresses to whatever addresses are
    > handed out through the DSL router. My PCs don't even know that the
    > addresses seen by the DSL are unrelated to the addresses known to them.
    >
    > This PC I'm working on wants to talk to the internet, so it sends a
    > request to an address that doesn't exist on my network out the wire. My
    > local router is the default route, so it gets the request because
    > there's no place on my network with that remote address. Because it's a
    > router, it routes the request out through its uplink to the DSL router.
    > It translates the PC address into the NAT address that it associated
    > with my PC, so any replies that come back will be tagged with the NAT
    > address. Replies come back tagged with the NAT address, and the router
    > knows it should translate that back to the original PC address before
    > putting the reply into my local network.
    >
    > Technically, I don't need to know anything about what the DSL router
    > does. I'm pretty sure that it's also doing its own NATting, so this PC
    > has its address translated twice. But it's really irrelevant to me. I
    > leave that up to the routers.
    >
    > In any case, my local router is what keeps the internet/DSL connection
    > alive. My PCs never "connect" to DSL. The DSL is always "on". From the
    > PCs' viewpoints (or my AS/400's), that's simply another network segment.
    >
    > As far as DNS goes, my ISP's DNS address gets inserted into my PCs right
    > after one gets powered on and starts talking to the network. I don't
    > really know what the real DNS address is; it's most likely NATted by the
    > DSL anyway. My PCs are actually using DHCP, but the requests are handled
    > by my local router and assigned within the local address range that I
    > chose. So far, the address for my ISP's DNS has never changed, so AFAIK
    > I could specify it as a static address.
    >
    > Because I use DHCP and the dynamic addresses are assigned by my local
    > router, the DNS address is also assigned by my router. However, it got
    > the address from DSL when it was powered on because I configured the
    > router to get its external address by DHCP from DSL.
    >
    > I turn the router on; it asks the DSL router for an address to use for
    > itself and an address for DNS. That DNS address gets copied to any PC
    > when the PC powers up.
    >
    > DNS and host tables have one particular use -- to allow you to use names
    > instead of IP addresses. If you never refer to a remote system by name,
    > you never need DNS nor host tables. A whole bunch of names get different
    > addresses at regular (or irregular) intervals. Because addresses change,
    > names are handy. DNS automates the association of name with address.
    >
    > For a local network, addresses can work fine. You don't really need host
    > tables on your PCs; they exist for convenience. I don't have a local
    > DNS, but I do have my AS/400 at a fixed (static) local address. I
    > usually store that name/address pair in a host table on my PCs, but
    > that's generally the only address in the PC host table.
    >
    > Well, that's not true. I actually have a bunch of addresses in this PC's
    > host table. Here are some of them:
    >
    > 127.0.0.1 a.tribalfusion.com
    > 127.0.0.1 ad.doubleclick.net
    > 127.0.0.1 servedby.advertising.com
    > 127.0.0.1 ad.targetingmarketplace.com
    > 127.0.0.1 media.fastclick.net
    > 127.0.0.1 adserver.00web.com
    > 127.0.0.1 anad.tacoda.net
    > 127.0.0.1 adopt.euroclick.com
    > 127.0.0.1 c.casalemedia.com
    >
    > You might recognize one or two.
    >
    > I collect them. They're representative of host names for many junk ads
    > that you see on lots of sites. When I find one that irritates me, I add
    > the complete host name to my hosts table with the localhost address
    > 127.0.0.1. When a web page refers to that host, the web browser looks at
    > the hosts table first. Since it finds a match, it sends the request to
    > the localhost address instead of whatever the real address is. The
    > localhost address has no clue what's being requested, so the request
    > gets rejected fairly quickly. Whatever image would have been sent never
    > really gets requested. The web page just has a blank spot there. The
    > pages show up a lot faster because there's less junk on it, and no
    > images have to take my bandwidth and no image gets copied into my cache.
    >
    > Sorry, for getting carried away with this response. The whole point is
    > that your PCs shouldn't have to "open" a connection through DSL except
    > when they're first turned on. The connection can simply "be there". It
    > can simply be available at all times.
    >
    > In my case, it's there because I kind of off-loaded that work to a
    > fairly cheap piece of hardware -- a cheap Linksys router.
    >
    > It's actually more complicated than that because (1) I have two local
    > network routers, (2) the other one is a bit more expensive than the
    > Linksys, (3) both of them are also wireless routers, (4) this PC has two
    > ethernet ports, one of which actually connects direct to the DSL router
    > and the other to my local router, (5) I have a KVM switch right here for
    > a second PC that has an ethernet port to my local router and also
    > connects via dialup to Earthlink, Netscape.com or Compuserve, (6) my
    > home network includes segments that are switched through the power lines
    > throughout my house, (7) four switches at different parts of the house
    > handle some small collections of devices...
    >
    > That is, once you get started experimenting, things just start sprouting
    > up all over. As I started this post out, it's all just experimenting
    > until it starts making sense. I'm hoping it'll make sense to me soon.
    >


  13. Re: Networking Questions

    Pat Barber wrote:

    > This entire tcp/ip networking thing has been a real
    > eye-opener for me. D.I.Y. networking is not my thing
    > but I will figure it out in time.


    I'm not much different. I buy some "obsolete" version of a piece of
    hardware for a few bucks every once in a while and start seeing if I can
    make it work. It's novice-level stuff that helps illuminate what I run
    into in real-world situations.

    > I much prefer the system side of things and I might also
    > say that SNA/SDLC makes tcp/ip look very crude in comparison.


    Very true. But most parts of both necessarily have counterparts in the
    other. Maybe that's how you should get your foundation built.

    For example, a MAC address is a MAC address either way. At that level,
    things look very similar. DNS servers have a kind of parallel with APPN
    network nodes (*NETNODE). An APPN network name kind of matches up with a
    TCP/IP domain name.

    We used to be able to register our APPN network names with IBM in order
    to facilitate internetworking among trading partners. That's somewhat
    similar to registering TCP/IP domain names. Host names in APPN and
    TCP/IP are both matched up with local network addresses and local
    networks are linked via network routing.

    A 5494 controller isn't conceptually much different from a router
    nowadays and a controller description would refer to it as a *LENNODE,
    kind of describing its Low-Entry Networking ability to link a small
    segment to a higher level network structure, e.g., to link to a *NETNODE
    that could route to other segments.

    A well-configured APPN network wasn't easy to find though. The business
    environment wasn't particularly willing simply to broadcast info that
    allowed easy remote connection without knowing the details of both ends.

    And it can be that little bit of broadcast info that can make all the
    difference. Connection info is generally made public for TCP/IP
    networking. Nobody would've proposed making exchange IDs, etc.,
    available via some public repository such as a world-wide DNS server
    network.

    But PCs got powerful. In order to make PCs interconnect without having
    every one of them attended by 'the network guy', some amount of info had
    to be made open to the rest of the world. The various protocols work the
    same no matter how big or small the device is. And all relevant info
    about how they work is also available. No need to locate an IBM problem
    analysis manual to interpret some obscure bind sense code. (And if you
    never had to delay a problem resolution because you couldn't get hold of
    a manual quickly, you were better prepared than I often was. Many sites
    had no manuals anywhere.)

    > Waaaay too many layers of obscurity, not to mention total
    > confusion.


    I went to just about every COMMON session on APPN/APPC for a number of
    years before I started understanding what the presenter was saying in
    the first 15 minutes of any session. In the beginning, I'd be as lost
    when I left as when I arrived. Once it started soaking in, it seemed
    "simple".

    I thought I knew it well enough until near the end of the old IBM
    FORUM/400 days. Then I got a very surprising awakening by running into
    Steve Glanstein on-line back in the early '90s. That's when I began
    realizing how many layers of obscurity SNA/APPN networking had.

    Most of the time, it wasn't relevant. We didn't need to know because we
    only had to connect through specifically configured connection points.
    Once I stuck a foot in the rabbit hole, I saw something of the
    underlying warren that made lots of alternative directions possible.

    If you know why FORUM/400 bit the dust, you have perspective on why
    TCP/IP is the standard. Not because it's exactly better, but because the
    vast majority of APPN networks and hosts were not properly configured
    and couldn't withstand open networking. No one wanted to delve into the
    levels that needed attention.

    > When I think back to the S/34 and just setting an address and
    > the system handled it makes this look pretty sad.


    True enough... until you wanted that system to talk to my system here in
    my house. Then we'd need to send each other the details necessary to
    create the remote descriptions and hope got them right. We likely
    wouldn't want them to be able simply to find each other and start
    talking. Not over a public network.

    In TCP/IP, it's not particularly different. You can assign an address
    and the system handles it. Or you can let the system assign whatever
    address is needed. That adds a piece of complexity if you choose to use
    it, but it's not required. In my case, that piece is handled by a cheap
    $50US router. That's cheap for the practical experience I gained.

    Also, with a little attention, you can see that TCP/IP configurations
    under OS/400 and i5/OS work just like they do under Windows. The GUI and
    green-screen interfaces tend to mask the similarities, but underneath
    the user interface, they're exactly the same.

    > I agree that TCP/IP is the new wave but what a giant leap
    > backward.


    Hey, I still use FTS! SNA/DS is also in regular use on systems I work
    with, for good reason. But you and I wouldn't be having this exchange
    without TCP/IP. And we mostly just expect it to 'work'. (And try
    picturing Google Earth via SNA.)

    As to how it manages to 'work', I'll stick with the thought of seeking
    out similarities between SNA & APPN and TCP/IP. As aspects match up in
    your mental map, more and more falls into place (IMO).

    As for your original problem, ask yourself why the users need to
    initiate a DSL connection at all. Why isn't it established at all times?
    Why invoke a dynamic process that needs to alter automatically
    system/communication configuration? Why not determine the proper initial
    configuration that can be set and used when needed?

    I'm not closely familiar with the exact process you originally
    described. You originally wrote "When you fire up the internet
    connect...", but I _never_ fire up any connect. It sounds as if they
    start the initial connection wizard every time rather than configuring
    once when the PC is first set up.

    I have seven PCs here plus an AS/400, all connected to DSL and no local
    DNS. My AS/400 is in the host table for some of the PCs, but I only use
    terminal or OpsNav access from a couple of them. Otherwise, I use
    Windows networking/Netserver for IFS access. All devices belong to the
    same Windows workgroup, so again no DNS.

    Not quite the size of your customer's site, but there are similarities.
    I don't need Exchange, Active Directory, SQL Server nor other
    complicating apps.

    Now, I use a DSL (proxy) router rather than a DSL modem. As such, I
    don't really need my personal router. I also have a DSL modem, but it's
    no longer in use. If I used the modem, I would definitely use my
    internal router because it would hold the userid/password to start up
    the DSL connection.

    Without that, the first PC to access DSL would have to tell the modem
    how to connect. I could then use a switch that all PCs would connect to
    and its uplink would connect to the modem. Once the modem connected
    outward, any other PC could use the existing connection through the switch.

    But it's easiest to load userid/password in a router and not use a
    switch. Let the router handle the connection, handle NAT, handle DHCP if
    used. No PC need be involved with that. As long as all devices are
    configured for the same subnet, everything should just 'work'.

    Come to think of it, maybe the subnetting is where the problem is. How
    clear are you on what subnetting IP ranges does? I should've asked about
    that first thing. As I look at earlier posts, I see a couple comments
    that indicate possible confusion.

    For example, you give:


    ....as a static address you might assign. Then:

    192.168.17.64
    255.255.255.0

    ....is an example of a DHCP address out of DSL. Those are essentially
    incompatible because (1) they assert that they're on different
    subnetworks and (2) any address of the form 192.168.x.x is
    "non-routable" by definition.

    Some groups of addresses are 'non-routable'. This is done in order to
    allow you to assign addresses within a local intranet that can't be
    directly accessed from another intranet. It's a degree of protection as
    well as a way of conserving addresses. Each local intranet can use the
    same addresses for its devices as are used in any other intranet in the
    world. There might be a million devices out there with an address of
    192.168.1.1 and you'd never know it.

    As a simplified overview, it goes kind of like this:

    192.168.1.1 host address
    binary is 11000000.10101000.00000001.00000001
    255.255.255.0 subnet "mask"
    binary is 11111111.11111111.11111111.00000000

    ANDing those together gives:
    network 11000000.10101000.00000001.00000000
    ....or 192.168.1.0 as the address of the network (subnet) that this host
    is a part of. The host address within that subnet is .1 or .00000001 in
    binary.

    The subnet "mask" is used to declare what "network" the host is a part
    of. By ANDing the mask and the IP address, you can separate the two
    elements -- network and host.

    But the other DSL-assigned address:

    192.168.17.64 host address
    binary is 11000000.10101000.00010001.01000000
    255.255.255.0 subnet "mask"
    binary is 11111111.11111111.11111111.00000000

    ANDing those together gives:
    network 11000000.10101000.00010001.00000000
    ....or 192.168.17.0 as the address of the network (subnet) that this host
    is a part of. The host address within that subnet is .64 or .01000000 in
    binary.

    Comparing the two network addresses, we know that 192.168.1.0 and
    192.168.17.0 are different networks. From TCP/IP networking definitions,
    we also can learn that addresses of the form 192.168.x.x are
    non-routable. Hosts on two different subnets with those addresses cannot
    communicate with each other (without help, e.g., NAT or some other
    technology). Even if they're physically on the same segment, they'll
    simply ignore each other. They're supposed to ignore each other.

    So, let's say you assign 192.168.1.100/255.255.255.0 to your AS/400. A
    PC with 192.168.1.1/255.255.255.0 has no problem establishing a
    connection. Both devices recognize that they're allow to talk.

    If the PC is changed to 192.168.17.64, both devices will suddenly refuse
    to talk. They've become foreign to each other. They're on different
    subnets with non-routable addresses. Not even a router will connect the
    two devices because of the non-routable addresses.

    So, the PC then sends a connection request to address 192.168.1.100.
    It's TCP/IP stack knows that that address is on a different subnet, so
    it sends the request to the default route, which would normally be a
    router. Of course, the router looks at the address and knows that it
    can't do anything with it, so it drops it. The PC simply gives up
    waiting for a response eventually.

    One _possible_ fix is to learn whether the DSL will always assign an
    address with subnet 192.168.17.0. If it does, then the basic answer is
    to assign the AS/400 with an address like 192.168.17.100.

    Of course, that will interfere with communication to any PC with an
    address like 192.168.1.1.

    So, now it comes down to knowing what addresses will be handed out by
    DSL. Maybe the addresses will range from 192.168.17.1 to 192.168.17.64.
    Or maybe they'll be 192.168.17.64 at the low end and .127 at the high
    end. It's almost a guarantee that the range won't cover the whole range
    from .1 to .254; there will be gaps that are defined.

    The gaps will be addresses that you can use as static addresses. As long
    as the network portion is compatible, determined by subnet masking, it
    shouldn't matter if addresses get dynamically assigned and static
    addresses get clobbered. Just assign static addresses that don't overlap
    the dynamic range and keep them in the same subnet.

    Maybe readdressing is all you need.

    --
    Tom Liotta
    http://zap.to/tl400

  14. Re: Networking Questions

    I think that putting the new box onto an existing
    switch with no knowledge of the addressing methods
    has made me crazy.

    I going back into the account this week and I
    will have a much greater knowledge of what they actually
    have there.

    I appreciate your comments..

    I'm going to attempt to find out what they currrently
    have and maybe shoot another post out on this...

    I have to read up on NAT today...



    Thomas wrote:


    > I'm not much different. I buy some "obsolete" version of a piece of
    > hardware for a few bucks every once in a while and start seeing if I can
    > make it work. It's novice-level stuff that helps illuminate what I run
    > into in real-world situations.


  15. Re: Networking Questions

    OK Tom, I did find out the "rest of the story"...

    DSL connection with AT&T

    Addresses are 12.103.237.194 ---12.103.237.206
    Router is 12.103.237.193
    Subnet 255.255.255.240
    Default Gateway is 12.103.237.193

    Apparently, they only have 13 addresses and that
    would explain several other problems...

    Assuming that all 13 are in use, that would screw
    any attempt at signing on to the Iseries would it
    not ???



    Pat Barber wrote:
    > I think that putting the new box onto an existing
    > switch with no knowledge of the addressing methods
    > has made me crazy.
    >
    > I going back into the account this week and I
    > will have a much greater knowledge of what they actually
    > have there.
    >
    > I appreciate your comments..
    >
    > I'm going to attempt to find out what they currrently
    > have and maybe shoot another post out on this...
    >
    > I have to read up on NAT today...
    >
    >
    >
    > Thomas wrote:
    >
    >
    >> I'm not much different. I buy some "obsolete" version of a piece of
    >> hardware for a few bucks every once in a while and start seeing if I
    >> can make it work. It's novice-level stuff that helps illuminate what I
    >> run into in real-world situations.


  16. Re: Networking Questions

    Pat Barber wrote:

    > DSL connection with AT&T
    >
    > Addresses are 12.103.237.194 ---12.103.237.206
    > Router is 12.103.237.193
    > Subnet 255.255.255.240
    > Default Gateway is 12.103.237.193


    Pat:

    Apologies for delay. I'm kind of stuck spending time except on weekends
    for a while.

    It sounds like you're getting a grasp on it.

    The subnet mask:

    255.255.255.240 binary 11111111.11111111.11111111.11110000

    The router:

    12.103.237.193 binary 00001100.1100111.11101101.11000001

    Therefore, the 'network' address is:

    12.103.237.192 binary 00001100.1100111.11101101.11000000
    00001100.1100111.11101101.1100xxxx

    The range of available addresses is determined by the low-order four
    bits according to the mask. The addresses may be 12.103.237.193 through
    12.103.237.207 where 207 is 192 + 15. We get "15" from the upper limit
    of binary 1111.

    Total number of addresses is 16 and one of those is the network address.
    One is the router and one for the AS/400. That leaves 13 addresses
    maximum left over to dole out via DHCP.

    I don't know how AT&T handles DHCP, whether you have a DSL modem or an
    actual DSL router that can be configured from your side. Questions to
    AT&T (or probably the ISP that provides it) may be in order.

    With only 13 available outside addresses, it seems to me you'll still
    want your own router. I don't think you want the AS/400 to have an
    exposed external address. (Technically, you don't want any addresses
    exposed. NAT is part of that.) By putting your router between your
    network and DSL, you have the first point of control separate from the
    outside. You'll trade the address of the AS/400 for the external address
    of the router, so it seems there'd still be only 13 addresses to use at
    one time to the outside.

    I'm not sure what more can be said at the moment. The change in your
    network is going to take some experience before you know what's going
    on. I think you've picked up enough to start recognizing things.

    Post obstacles as they come up. Maybe someone with more networking
    expertise than I've got will add more.

    --
    Tom Liotta
    http://zap.to/tl400

  17. Re: Networking Questions

    The new line, the new router and the new firewall all
    should show up pretty soon. Once we get it all installed
    I'll be back with more questions.

    I'm slowly beginning to see the light(maybe).

    Thanks for your responses.


    Thomas wrote:

    >
    > Pat:
    >
    > Apologies for delay. I'm kind of stuck spending time except on weekends
    > for a while.
    >
    > It sounds like you're getting a grasp on it.


+ Reply to Thread