Gang Li & Yun Huang |
· Network Adapter Driver for ESCON
· MPC+ for Windows NT
· TCP/IP Direct Connect
· Data Link Control for
Communications Server for Windows NT
The network adapter driver provides a standard windows NT adapter driver for the PCI ESCON adapter. This driver is installed the same as any other network interface connection through the windows NT control panel.
MPC+ for windows NT provides the necessary protocol to communicate with the OS/390 system using multipath channel+. MPC+ is IBM's high-performance channel protocol specifically designed for networking. The MPC+ for windows NT software supports both TCP/IP and APPN(SNA) links to the mainframe.
TCP/IP direct connect provides the ESCON adapter a network driver interface specification (NDIS) compliant interface. Microsoft's TCP/IP protocol communicates with network adapters (Ethernet, Token Ring, FDDI, etc.) using the NDIS interface. Therefore, once TCP/IP direct connect is installed and configured, the server's TCP/IP stack can communicate with the ESCON adapter in the same fashion it uses to communicate with any other network interface.
Data link control provides the necessary drivers to connect MPC+ and the
ESCON adapter to communications server for windows NT.
The supported releases of microsoft software are:
-
Windows NT Server version 4.0 Service Pack 5
OS/390
When using a Channel-to-Channel TCP/IP link with TCP/IP direct connect the
following minimum operating system release is required:
-
OS/390
DataBase
The following releases of database software are supported:
- IBM DB2 for
OS/390 version 5.1 or above
- IBM DB2 for MVS/ESA version 3.1 or 4.1 (for
APPC connectivity only)
DRDA Application Requester Software
The following releases of software are supported on windows NT:
- IBM DB2
Connect Enterprise Edition version 5.2 or above
- IBM DB2 Connect Unlimited
Edition version 6.1
- IBM DB2 Universal Data Base - Enterprise Edition
version 5.2 or above
Communications Software
- IBM communications server for windows NT version 6.02 with APAR JR 14007 or above (only required for APPN communications)
Step
Installing the Network Adapter Driver
Once the adapter is installed into the server the next step is to install the windows NT network adapter driver. This driver is always required in order to allow windows NT to communicate with the adapter. The driver must be installed once for each PCI ESCON Adapter installed in the system. The following step-by-step procedure will install the driver into your windows NT server.
Step
The network adapter driver must be installed before you begin the installation of MPC+ for Windows NT. Once you have completed the network adapter driver install and re-booted the server the following procedure will install the MPC+ software.
Step
TCP/IP Direct Connect is installed like any other network adapter driver. TCP/IP Direct Connect must be installed once for each MPC transmission group you intend to define between the server and a mainframe.
Step
1. Insert the TCP/IP direct connect
diskette that came with MPC+ for windows NT into your floppy drive. From the
windows NT
control panel invoke network configuration.
2. Click on the Adapters tab.
3.
Click the Add button to add a new adapter. The select network
adapter dialog box will display a list of adapters
included with
windows NT.
4. Click the Have
Disk button.
5. Click OK.
6. The Select OEM Option dialog will
display TCP/IP direct connect. Click OK to install the software.
7. The TCP/IP Direct connect properties
screen for channel to channel (CTC) link type will display.
8. Both HSAS and MPCPTP are MPC+ links.
MPCPTP is used for DB2 connect TCP/IP links. Click the MPCPTP
radio button.
The TCP/IP direct connect properties screen will display MPC+ configuration
fields.
Complete the following fields:
Mainframe IP Address - The IP address of the
mainframe. This address ust match the source IP
address
specified on the HOME IP address defined in the TCP/IP profile.
Link Description - An informational link description
MPC Group - Identifies which MPC transmission group
this link is to use. You must select a group
from the
list of available groups. If no groups are available this pull-down list will be
empty. To define
or modify a group, click the
Configure MPC... button.
MAXBFRU - The maximum number of 4096 byte buffers
to be used. For HSAS it must be set to 9.
For MPCPTP 16
is recommended. Note the TRL definition on the mainframe must match.
MTU - Enter the maximum transmission unit as it
appears on your oeifconfig statement for HSAS or
the MTU
size defined in the TCP/IP profile for MPCPYP.
9. Click Close to
save the configuration and return to the Network Adapters list.
10. Click on Bindings. Show bindings
for all adapters. Check that only the TCP/IP protocol is bound to the TCP/IP
direct
connect adapter. Disable any other service or protocol bound to TCP/IP direct
connect.
11. Click Close to end network
configuration. Windows NT will perform an automatic bind process. TCP/IP direct
connect
is bound to Microsoft's TCP/IP stack on the server. When windows NT attempts to
perform this bind, the
standard
Microsoft TCP/IP properties dialog box will display.
12. Use your mouse to select the TCP/IP direct connect adapter from
the list of adapters bound to TCP/IP.
13. Enter an IP address and subnet mask; defining the address
that the mainframe will use to communicate with the server.
Note that the mainframe IP address and this address must be in the same network
(subnet) and that subnet must be
different
from other subnets used with network interface cards (NICs) in the server. There
is no need to enter a default
gateway since TCP/IP direct connect is a point-to-point connection.
14. Click OK. Windows NT will complete the
binding process and prompt you to reboot the server. You must reboot for
network
changes to take affect.
Device Address- The Sub-Channel address for the device, ranging in value from 00-FF hexadecimal. This address should match that coded in the UNITADD parameter of the CNTLUNIT macro in the IOCDS. Sub-channel addresses must be unique across any one partition.
For MPC+ two or more unit addresses must be defined. These device addresses must be coded as UNIT=SCTC.
EmulationType - The type of IBM controller this device is to emulate to the mainframe. Set to "MPC" for TCP/IP Direct connect and MPC+ for windows NT.
Control Unit - The control unit address the mainframe will use to communicate with the controller and the controller will use to identify itself to the mainframe. Control unit address is a hexadecimal number ranging from 0-F. This number must match that coded in the CUADD parameter of the CNTLUNIT macro in the IOCDS. Note that CUADD is an optional field in the IOCDS and defaults to zero if not coded.
Channel Link - The address of the ESCON director port that connects to the mainframe this device is to communicate with. The value range is from 01-FE hexadecimal. The mainframe uses a channel link address '02' to communicate with Server 1. Server 1 uses a channel link address '01' to communicate with the mainframe. In this example, the channel link in the adapter configuration is set to '01'.
If no director is used, the channel sub-system will assign an arbitrary LINK address for the server to communicate with the mainframe. To ensure the mainframe always selects '01' you should assign '02' to the channel adapter's device in the IOCDS. Code the LINK parameter in the CNTLUNIT macro in the I/O configuration data set to '02' as shown in the following example:
CNTLUNIT ..., LINK=(02),...
Then set the channel link '01' in the
adapter's configuration.
Logical Partition Address (LPARS) - This is a hexadecimal number ranging from 0-F identifying the logical partition ID of the Logical partition this connection is to communicate with. This parameter is only required if EMIF is enabled. For BASIC mode or LPAR mode non-EMIF, set this field to zero (0). If EMIF is enabled this number must match the LPAR ID assigned in the Logical Partition Definition (LPDEF) frame which appears in LPAR mode after a power on/reset. In LPAR mode the CPU ID contains this number as it's fourth character. If running the VM operating system, the "QUERY CPUID" command will display the CPU ID. The fourth character of the CPU ID is the LPAR ID.
ESCON configuration parameters are viewed and set through the windows NT system control panel. From the main screen of control panel double click the Network Icon.
Select the Adapters tab.
The Bus-Tech PCI ESCON Adapter should be included in the list.
Selecting the Bus-Tech adapter and clicking Properties will bring up the ESCON configuration screen.
To configure a new device click Add ...
Complete the channel link address, LPAR,CUAdd, Base device address and emulation type fields in the Add ESCON Devices dialog according to the descriptions earlier in this chapter. If your mainframe is configured with ESCON multiple image facility (EMIF) and this CHPID is defined on the mainframe as shared, then check the CHPID is shared check box. Enter the number of devices you wish to define. For example, to create device A0 and A1, enter2. Optionally enter a Description to be associated with this device.
Press Add to add the device to the ESCON configuration file. The ESCON Adapter properties dialog will re-display, now showing the new devices.
To remove a previously defined device, highlight the device and then click Remove. To edit a previously defined device, highlight the device and then click Modify.
To save a newly created configuration, click OK.
Step
1. From the MPC configuration screen,
click the Add... button.
2. The MPC
group Add/Edit screen is used to define a new transmission group. Enter a group
name and an informational
description.
3. Each MPC group requires at least one
subchannel address assigned for Host Read Connections and one subchannel
address
assigned for Host Write Connections.
To add a subchannel to either the Host Read Connections list or the Host Write Connections list, click the Add... button under that list. The Add connection screen displays.
4. Select the ESCON adapter you are
using and enter a device address. Make sure to use an address defined in the
adapter's ESCON
configuration as "MPC". Click OK.
5. Once both Read and Write connections have been defined, click OK
to define the group and return to the MPC
configuration
screen which now shows the new MPC group.
6. Click OK to complete the MPC transmission group configuration.
At least two sub-channels must be defined for each transmission group; one
for read activity and one for write activity. Devices defined for a MPC
transmission group must be defined as serial channel to channel devices in the
mainframe IOCP.
More than two devices may be defined. Devices may be on
different ESCON paths. More specifically, if your windows NT server has more
than one ESCON adapter installed you may define devices on different adapters
into the same MPC group to increase overall throughput and provide redundancy in
event of a path failure.
OS/390 version 2 release 4 and above uses the HCD
configuration utility, a menubased interactive program to complete this task.
Refer to the OS/390 HCD user's guide for instructions on using HCD.
In HCD
define devices with Device Type "SCTC".
Updating the OS/390 Parmlib
IBM recommends that the missing interrupt handling processes be turned off
for devices being used for MPC+. To accomplish this add the devices to OS/390's
SYS1.PARMLIB(IECIOSxx) as follows:
MIH TIME=00:00, DEV=(xxx-yyy)
where
xxx-yyy is the range for devices you intend to use for MPC+; for example
200-201.
Modify the TCP/IP Profile
The following bolded lines provide a sample of the control statements that
must be added to the TCP/IP profile in order to define a new MPC link.
| ; = = = = = = = = = = =
= MPC Profile Entries = == = = = = = = = = =
= =
DEVICE
MPCGP1 MPCPTP
HOME GATEWAY |
The EnterpriseExpress adapter for DB2 access can be used with both DB2 connect Enterprise Edition or DB2 connect personal edition. Verification of the TCP/IP link between the windows NT server and the mainframe can always be performed from the console of the NT server. However, if you are using DB2 connect personal edition it is best to verify the link from the client to insure IP forwarding is properly configured in the server. Please refer to the appropriate section below.
DB2 Connect Enterprise Edition
As shown in below Figure, DB2 connect communicate with the ESCON adapter
using TCP/IP and TCP/IP Direct connect(NDIS). TCP/IP direct connect uses MPC+
across the channel to communicate with the Mainframe.
| ESCON Adapter |
| MPC+ for Windows NT |
| TCP/IP Direct Connect |
|
Windows NT
TCP/IP Stack |
| DB2 Connect Enterprise Edition |
Once you have completed installing an configuring the adapter as described in
the previous chapters of this manual, you can easily verify the TCP/IP
connection to the mainframe with the Windows NT PING command.
At the console
of our NT Server, open a MS-DOS command prompt and type in a PING command using
the IP address assigned to the mainframe in the TCP/IP profile.
If the ESCON
connection is properly installed and configured, your mainframe's IP stack will
respond to the PING(below Figure).
| Microsoft<R> Windows NT<TM>
<C> Copyright 1985-1996 Microsoft Corp. C:\> Ping 207.221.147.65 Pinging 207.221.147.65 with 32 bytes of data: Reply from 207.221.147.65:
bytes=32 time<10ms TTL=128 C:\>
|
ESCON CTC Support for Parallel CTC Basic Mode
The parallel interface architecture described in IBM Channel-to-Channel
Adapter, SA22-7091, defines both the basic and extended CTC operating modes. The
ESCON interface described in ESA/390 ESCON Channel-to-Channel Adapter,
SA22-7203, defines only one operating mode (extended mode). This operating
mode is similar to the extended mode for the parallel interface described in
IBM Channel-to-Channel Adapter, but with changes that support the ESA/390
ESCON I/O interface architecture. See IBM Channel-to-Channel Adapter and
ESA/390 ESCON I/O Interface, SA22-7202 for more information.
The architecture described in ESA/390 ESCON Channel-to-Channel Adapter does not support CTC basic mode.
However, the following CPCs provide CTC basic mode support using ESCON CTC channels:
2003 models (all ECs)
9672 models (all ECs)
9021 711-based models (all SECs)
9021 520-based models at SEC 228112
9121 511-based models at SEC C22840
The implementation is based on the basic mode section of the parallel
interface architecture described in IBM Channel-to-Channel Adapter, with minor
modifications for adapting to the ESCON I/O interface.
The CTC basic mode support is invoked, on a device basis, by coding BCTC
instead of SCTC for the UNIT keyword on the IODEVICE macro. The default
device type for an SCTC control unit is SCTC, which specifies the device is
to operate in the mode defined in the ESCON CTC architecture.
There is no change to the CNTLUNIT macro when invoking basic mode. Both SCTC
and BCTC devices can be defined to the same CTC control unit with
keyword
UNIT=SCTC in the CNTLUNIT macro.
The following IODEVICE macros define one BCTC device (device number= 5150)
and three SCTC devices (device numbers 5151-5153) to an unshared CTC
channel.
CNTLUNIT
CUNUMBR=5150,PATH=51,LINK=D0,CUADD=5,UNITADD=((40,4)),UNIT=SCTC
IODEVICE
ADDRESS=(5150,1),CUNUMBR=5150,UNITADD=40,UNIT=BCTC
IODEVICE ADDRESS=(5151,3),CUNUMBR=5150,UNITADD=41,UNIT=SCTC
The following IODEVICE macros define, to a shared CTC channel:
One BCTC device (device number=5150)
Three SCTC devices (device numbers 5151-5153) for LP1A to LP5B
Three SCTC devices (device numbers 5151-5153) for LP2A to LP5B
Four SCTC devices (device numbers 5151-5154) for
LP3A to LP5B
A total of 15 CTC devices (5 devices for each of the 3 pseudo control units) are allocated toward the 512 CTC devices support limit.
CNTLUNIT
CUNUMBR=5150,PATH=51,LINK=D0,CUADD=5,UNITADD=((40,5)),UNIT=SCTC
IODEVICE
ADDRESS=(5150,1),CUNUMBR=5150,UNITADD=40,PART=(LP1A),UNIT=BCTC
IODEVICE
ADDRESS=(5151,3),CUNUMBR=5150,UNITADD=41,PART=(LP1A,LP2A,LP3A),UNIT=SCTC
IODEVICE
ADDRESS=(5154,1),CUNUMBR=5150,UNITADD=44,PART=(LP3A),UNIT=SCTC
Unlike the IBM 3088 CTC devices, once an ESCON CTC device operating mode is
specified by an IODEVICE macro, it cannot be switched to the other mode
by
the CTC command. The operating mode can only be changed by re-configuring the
CTC device through dynamic I/O by means of MVS/HCD or a system IML
with a
new IOCDS. To change a device's operating mode, MVS/HCD must delete the device
first, then add the same device back in with a different mode.
Differences from Parallel I/O CTC Basic Mode
The SCTC devices perform in compliance with the ESCON CTC architecture. The
BCTC devices perform in accordance with the basic mode section of parallel
I/O CTC Architecture with the following deviation:
1.If the CCW byte counts are equal for read/write matching pair
commands pending on opposite sides of an ESCON CTC device (either SCTC or BCTC),
neither channel detects an incorrect-length
condition.
2.The ESCON BCTC devices cannot be switched to the extended mode
by the Set Extended Mode CCW command. Set Extend Mode command is now
interpreted as No-op. The following command codes,
which are assigned by the parallel I/O basic mode CTC architecture, are now
interpreted as No-op:
11001011, 1101x011, 11101011,
and 1111x011.
3.The test I/O (TIO) command is not supported in the ESCON I/O
interface architecture. Thus, it is not supported by the ESCON BCTC devices. If
the TIO
command is received by a BCTC device, it
causes the CTC channel to detect an interface-control-check (IFCC) condition.
4.Data chaining in CTC channel programs is not supported. If the chain data CCW flag is set, the BCTC device detects an IFCC condition.
The Sense ID (SID) command response is used by programs to verify the device
operating mode. The SID response data for a BCTC device, as shown below, is
the same for an SCTC device except that the device model number (byte 6) is
X'01' for BCTC devices and X'00' for SCTC devices.
Byte 0:
X'FF' Byte 7: X'00'
Byte 1:
X'30' Byte 8: X'40'
Byte 2:
X'88' Byte 9: X'C4'
Byte 3:
X'1F' Byte 10: X'00'
Byte 4:
X'00' Byte 11: X'60'
Byte 5: X'00'
Byte 6:
X'01'
Read Configuration Data Command
A new Read Configuration Data (RCD) CCW command is added with command code
X'11000100'.
The node element descriptor (NED) in the
configuration record of the RCD command response can be used by programs to
verify the device operating mode. Bytes 10-12 of the NED indicate the EBCDIC
characters CTB (X'C3 E3 C2') for BCTC devices and the EBCDIC characters CTC
(X'C3 E3 C3') for SCTC devices.
The following deviations from the ESCON I/O interface architecture are necessary for software compatibility with existing CTC basic mode programs:
The parallel I/O CTC basic mode architecture does
not support unit-check status for reporting errors. Instead, it uses the
'disconnect-in' sequence to cause
the attached
parallel channel to detect an interface-control-check condition. Because there
is no equivalent of the parallel I/O 'disconnect-in' sequence in the
ESCON I/O architecture, a BCTC device can no longer
report detected errors (link errors, hardware errors) by means of this
mechanism. Instead, it uses the
unconditional-disconnect (UD) / unconditional-disconnect-response (UDR) sequence
of the ESCON I/O architecture to cause the attached ESCON channel
to detect an interface-control-check (IFCC)
condition. The BCTC device does not present unit check status following
completion of the UD/UDR sequence
as it is required
by the ESCON I/O architecture. It depends on the channel to detect an IFCC
error.
Sense CCW command defined by the ESCON I/O architecture is not supported.
The resetting-event condition defined by the ESCON I/O architecture is not supported.
In basic mode X'04' command equals sense command
byte (SCB).