Adding SNASW layer to existing DLSW network

[admin_2]

Have an existing DLSW+ network, supporting PUT2.0's through IBM 3745's to Mainframe. Mainframe had been upgraded to Z-Series with OSA-Express adapter and APPN. Will be adding a new layer of SNASW routers between the DLSW+ Host end routers and the OSA-Express adapter. New connectivity path will be from end site router, sdlc serial attached PUT2.0 with vmac, encap into DLSW thru WAN core to DLSW+ Host Routers (2 7206's) at enterprise data center, then over crossconnect ethernet into 2 new SNASW 7206's, then into OSA-Express adapter and APPN/HPR-IP. End Site router's DLSW is configured with the two DLSW remote-peers and dlsw round robin for load distribution and redundancy.

FROM SNASW DOWNSTREAM>

Since the DLSW+ host router terminates the DLSW peer sessions, then a SNA datagram LLC should continue it's flow upstream to the directly connected, next hop SNASW router looking for the SNASW's mac address configured on the SNASW PORT statement. (Q1) How is the port configured between the SNASW router and the DLSW+ Host router? As a SNASW PORT portname VDLC or something other than VDLC? Do I also need SAP type and CONNTYPE of NOHPR? (just planning on DLSW puT2.0's, no BX or EE end nodes yet.)

FROM SNASW UPSTREAM>

Both SNA switching routers will use EE function (UDP/IP) upstream to OSA-Express Adapter. APPN with HPR-IP support is on mainframe. TCPIP for MVS is also used. The SNASW PORT portname HPR-IP FASTHETHERNET0/1 statement suggests that it will re-encapsulate the sna datagram llc into UDP/IP, then send it over the FastEthernet0/1 using FE0/1's IP address as the source address, on to the OSA-Express and the APPN HPR-IP destination. Since I will have two SNASW routers in place, each will have a unique IP address so that APPN will know which SNASW router to return the reply to.

(Q2) Does this sound like a proper design for using dual SNASW routers upstream?

(Q3) Any configuration statement recommendations?

[romney]

Hi Robert.

(A1) VDLC is only used when you have SNASw and DLSW on the same router. In your case you will associate the downstream snasw port with the Ethernet interface(s). You will have to use separate VLANs for duplicate destination mac addresses (and you may optionally choose to use HSRP for backup within each Ethernet VLAN). You will not need to specifiy a SAP (unless you are using HSRP and have multiple links from the same remote mac address). You can specify CONNTYPE NOHPR to guarantee no downstream devices can connect as HPR capable, but that may not be necessary (it depends on how the downstream devices are configured).

(A2) Yes, EE upstream as you describe is fine. The SNA data is sent upstream with a UDP header. The hpr-ip link does not use llc, but instead uses a logical data link control (ldlc - more in A3).

(A3) I recommend you configure ldlc on the snasw hpr-ip port so that its values match those on the VTAM hpr-ip port (the ldlc keyword has three values which match the LIVTIME, SRQTIME, and SRQRETRY parameters on the M/F). Many customers choose to define a loopback interface to associate the snasw hpr-ip port with, but there is no problem with using a real interface. You will also need to define a link to your preferred NNS and a backup NNS (if there is one). And you will need links to any other LPARs you wish to have a direct connection with, or you can associate a vnname on the hpr-ip port(s) and have dynamic connection network links created on-demand (you also need the same VNNAME defined on each LPAR).

Please let me know if you have any additional questions.

- Ray

[#3]

Our SNASW routers will physically stack on top of our DLSW host routers, so in lieu of VLANs and a switch, we plan to use direct crossover ethernet cables between an SNASw router and a DLSW host router. I think that using the same mac address on the two SNASW routers will be fine, as the DLSW remote peer and round robin process will help the end site router select a particular DLSW host router, each of which defaults by direct connection to a specific SNASW router which embodies the target mac address. Each SNASW router will then have unique IP addresses on the upstream ports to the OSA-Express adapter.

(Q4) I'm curious if the target M/F destination ip address, specified in the SNASW LINK statement is an IP address in the M/F APPN config or the IP Address in the M/F's TCP/IP profile (e.g. VIPA)?

(Q5) None of the SNASW command references that I've accessed shows the ldlc parameter as an option. Also haven't found the M/F side of the LIVTIME, SRQTIME, and SRQRETRY parameters.

[romney]

Sorry I jumped to the VLAN conclusion ... crossover ethernet to connect is fine.

(A4) You specify the ip address you have defined on the IPADDR start option on the target M/F. If your M/F doesn't have IPADDR specified it defaults to the first VIPA address in the TCP/IP stack’s (the referenced TCPNAME start option stack) HOME list.

(A5) The ldlc parameter was added via ddts, and hasn't made it to the publications yet. When you are in configuration mode you will see it when you enter snasw port EE hpr-ip Ethernet 0 ?. In the z/390 Resource Definition Reference you can find LIVTIME, SRQTIME, and SRQRETRY in Chapter 2, Major Nodes, External Communication Adapter (XCA) Major Node, Enterprise Extender Connections. You can look in your VTAMLST for the XCA major node containing your hpr-ip port to see what you currently have configured for these operands.

[#5]

I like your suggestion on using a loopback interface for the SNASW PORT HPRIP statement. I guess I would define the unique IP address as loopback 1? How does the loopback1 for the SNASW PORT HPRIP get associated with the physical FastEthernet0/1 that will connect upstream to the customers OSA-Express adapter? Does this FastEthernet need an ip address or just a update source loopback 1 statement?

[#6]

I decided to just use the ethernet interface for now and not the loopback>

I cannot find any references to SNASW commands that have to correlate with the APPN HPR EE definitions on the main frame. It appears that the SNASW LINK name, gets passed up to the host, and the host has to recognize this name by explicitly defining it in order to authenticate the bind request. May have to call TAC.

[htranphu]

Robert,

You can configure a SNASw port for the loopback (the loopback has a unique IP address), and then the SNASw upstream link using that port, and pointing to the IP address of the host. Between SNASw and the host, we will then use IP routing; so which physical interface is used depends on your IP network in between.

This is an extract of the SNASw config:

...

interface Loopback0

ip address 10.20.10.10 255.255.255.0

...

snasw port EE hpr-ip Loopback0

snasw link MVSDEE port EE ip-dest 172.18.1.41

I hope this helps.

tran

[romney · Accepted solution]

Robert,

Sorry to be so slow to respond. I see Tran answered your loopback question if you decide to use it in the future.

With regard to the snasw link activation, you either have to have dynamic PU support (DYNPU=YES) indicated on the host HPR-IP GROUP, or you must predefine the PU in a switched major node with a NETID/CPNAME that matches the snasw cpname configured on the router. The snasw link name is passed to the host, but it is usually the snasw cpname (from the xid) that VTAM uses to find a predefined EE PU. I'm not sure where you were going with your reference to the BIND request, but if we haven't answered your question please further clarify.

Below I've included our host xca major node for EE. We have DYNPU=YES, so there is no need to predefine the snasw 2.1 adjacent link station (PU).

- Ray

EEXCAJ VBUILD TYPE=XCA

EETGJ PORT MEDIUM=HPRIP, X

VNNAME=EEJEB, X

VNGROUP=EEGRPJ, X

LIVTIME=10, X

SRQTIME=15, X

SRQRETRY=3, X

SAPADDR=04

*

EEGRPJ GROUP ANSWER=ON, X

AUTOGEN=(254,L,P), X

CALL=INOUT, X

DIAL=YES, X

DYNPU=YES, X

DYNPUPFX=$E, X

ISTATUS=ACTIVE