G VSM4 FICON Front-End and Back-End Configuration

The VSM4 FICON Back-End connectivity feature adds value to the previously available FICON front-end connectivity. Table G-1 summarizes the supported card configurations for VSM4 FICON Front-End plus Back-End connectivity.

Table G-1 Supported Card Configurations for VSM4 FICON Front-End plus Back-End Connectivity

VCF Cards	FICON Ports	ICE Cards	ESCON Ports	Total Ports	Total Logical Paths (16 per ICE Port, 64 per VCF Port)
2	4	6	24	28	640
4	8	4	16	24	768
6	12	2	8	20	896
8	16	0	0	16	1024

VSM4 FICON VCF Card Options

VSM4 supports the following FICON VCF card options:

Figure G-1 shows a VSM4 with 6 ICE cards, 2 VCF cards.
Figure G-2 shows a VSM4 with 4 ICE cards, 4 VCF cards.
Figure G-3 shows a VSM4 with 2 ICE cards, 6 VCF cards.
Figure G-4 shows a VSM4 with 8 VCF cards.

Figure G-1 VSM4 with 6 ICE Cards, 2 VCF Cards

Figure G-2 VSM4 with 4 ICE Cards, 4 VCF Cards

Figure G-3 VSM4 with 2 ICE Cards, 6 VCF Cards

Figure G-4 VSM4 with 8 VCF Cards

Note:

In Figure G-1 through Figure G-4, the VCF cards must go in:

Slots 5 and 15 in a two-VCF card configuration
Slots 3, 5, 13, and 15 in a four-VCF card configuration.
Slots 3, 4, 5, 13, 14, and 15 in a six-VCF card configuration.
All slots in an eight-VCF card configuration.

FICON ports are controlled by a FICON Interface processor (FIP), ESCON ports are controlled by a CIP. Regardless of the card configuration, there can be only a total of 14 Nearlink FIPs and/or CIPs.

With microcode level D02.06.00.00 or higher, multiple Nearlink device connections through a FICON or ESCON switch or Director on the same port now allow:
- Up to a total of 16 simultaneous NearLink I/O transfers, which can be spread across multiple targets on as many as 14 NearLink ports.
- Up to a total of 2 simultaneous NearLink I/O transfers are allowed per port.
All FICON ports can be configured as either a Host port or Nearlink (RTD/CLINK origination) port. All ESCON ports continue to be configurable as host or Nearlink ports in pairs on a per CIP basis.
As shown in Figure G-1 through Figure G-4, the ports are shown with their channel interface identifiers where all ports are enabled. These channel interface identifiers are the values that are required for the CHANIF values that you code for the CONFIG utility. Each value is two characters in length and has a value from 0A to 1O. The first digit is the VTSS cluster ID (valid values are 0 or 1). The second digit is the group or adapter ID (valid values are A to P).

Each FICON port can attach to two RTDs, or two CLINKs, or an RTD/CLINK combination using a FICON director or supported switch (in FICON mode). Note that, as shown in these figures, for RTDs only, each FICON port has two CHANIF values only if the port is connected to a FICON director which is then connected to two RTDs. Nearlink RTD connections that are paired through a FICON switch or director on the same port dynamically alternate between both RTDs for atomic operations such as mount, migrate VTV, recall VTV, and so forth.
Each ICE card contains two pairs of ESCON ports. Each pair is controlled by its own Channel Interface Processor (CIP). Each CIP switches between the two ports, so that only one port can transfer data at a time, which emulates a FICON port attached to a director attached to RTDs.
Each host FICON channel supports 64 logical paths (times 16 logical units). However, in HCD:
- From a single MVS host, you can only define 8 channels (CHPIDs) running to a single control unit (single VSM4).
- You use the CNTLUNIT statement to define each VSM4 as 16 3490 control unit images.
- You use the IODEVICE statement to define the 16 VTDs that are associated with each 3490 control unit image.
For a VSM4, each ESCON CIP or FICON FIP can operate with only one of two modes, which is set at the VTSS LOP:
- Host Mode. In Host Mode, ports can connect to the host CPU channels, including using Director(s) or channel extenders. A port in Host Mode can also serve as a CLINK terminator.
  
  Also note that for ESCON ports, you can have two physical paths from the same LPAR to the same CIP, as long as the two physical paths address different (not overlapping) logical control units. For example, a single host LPAR can address logical control units 0-7 on one CIP port, and 8-F on the other CIP port of the same CIP.
- Nearlink Mode. In Nearlink Mode, ports can connect to an RTD. A port in Nearlink Mode can also serve as a CLINK originator.
- For clustering, you need an originator port in Nearlink mode on one VTSS connected through a CLINK to a terminator port in Host mode on the other VTSS.
  
  Caution:
  In bi-directional clustering, each CLINK must be attached to the same Storage Cluster on each VTSS, which is a requirement. Failure to configure in this manner can produce Replicate, Channel, and Communication errors. For more information and examples, refer to ELS Disaster Recovery and Offsite Data Management Guide.

FICON/ESCON Best Practices

For both FICON and ESCON, Best Practices for optimizing port operations are shown in Table G-2.

Table G-2 Optimizing VSM4 FICON/ESCON Port Operations

Configuration - Two ESCON Ports on a CIP (ICE) or FICON port attached to a FICON Director (VCF)	Best Practices
Two CLINKs	Attach a maximum of 2 because each port allows two active operations. Note, however, that these operations share the bandwidth of the port.
CLINK and RTD	An advantage if you attach one CLINK originator/one RTD per director, because both can be active.
Two RTDs	An advantage for the following: Optimize use of local and remote RTDs. During busy shifts, use only local RTDs on the FIP. During quiet periods, switch to remote RTDs for deep archive and DR work. Because you can have two active devices, you can also simultaneously run one local and one remote RTD. Note, however, that these operations share the bandwidth of the port. Optimize use of different drive technologies. As described in the previous bullet, use a T9840 as a local RTD, then switch to a T9940 for deep archive. You can also use this feature to migrate from older drive technology (such as 9490) to newer technology (such as 9840). Use Management and Storage Classes to read in data from older media, then switch to the newer technology drive to place data on new media. This technique effectively gives you greater physical connectivity to different drive technologies without incurring the overhead of full time, real time FICON connections to each drive type. Also as above, because you can have two active devices, you can also simultaneously run two RTDs with different drive technologies. Note, however, that these operations share the bandwidth of the port.

Configuration - Two ESCON Ports on a CIP (ICE) or FICON port attached to a FICON Director (VCF)

Best Practices

Two CLINKs

Attach a maximum of 2 because each port allows two active operations. Note, however, that these operations share the bandwidth of the port.

CLINK and RTD

An advantage if you attach one CLINK originator/one RTD per director, because both can be active.

Two RTDs

An advantage for the following:

Optimize use of local and remote RTDs. During busy shifts, use only local RTDs on the FIP. During quiet periods, switch to remote RTDs for deep archive and DR work. Because you can have two active devices, you can also simultaneously run one local and one remote RTD. Note, however, that these operations share the bandwidth of the port.
Optimize use of different drive technologies. As described in the previous bullet, use a T9840 as a local RTD, then switch to a T9940 for deep archive. You can also use this feature to migrate from older drive technology (such as 9490) to newer technology (such as 9840). Use Management and Storage Classes to read in data from older media, then switch to the newer technology drive to place data on new media. This technique effectively gives you greater physical connectivity to different drive technologies without incurring the overhead of full time, real time FICON connections to each drive type. Also as above, because you can have two active devices, you can also simultaneously run two RTDs with different drive technologies. Note, however, that these operations share the bandwidth of the port.

VSM4 FICON Front-End and Back-End Configuration Examples

For VSM4s with both FICON Front-End and Back-End connectivity, look at two examples of VCF card configurations and implementation:

"VSM4 Configuration Example: 8 VCF Cards, FICON Directors, 16 RTDs"
"VSM4 Configuration Example: 8 VCF Cards, 4 CLINKs, FICON Directors for 8 RTDs"

For a VSM4 host gen example, see "IOCP Example for Single MVS Host Connected to a VSM4 Using FICON Directors."

VSM4 Configuration Example: 8 VCF Cards, FICON Directors, 16 RTDs

Figure G-5 shows CONFIG channel interface identifiers for a VSM4 with 8 VCF cards. In this configuration, you have allocated 8 ports to RTDs and 8 ports to host connections. The RTD ports are all connected to FICON directors, each of which is attached to RTDs, so the CHANIF identifiers for both RTDs are shown on each port. This allows Back-End connection to 16 RTDs, although, as with ESCON, only one RTD per port/Director can be active at a time.

Figure G-5 VSM4 with 8 VCF Cards, FICON Directors for 16 RTDs

CONFIG Example for VSM4 FICON with 8 VCF Cards, FICON Directors, 16 RTDs

The following example shows CONFIG JCL to define the VSM4 configuration in Figure G-5.

//CREATECF EXEC PGM=SLUADMIN,PARM='MIXED'
//STEPLIB DD DSN=hlq.SEALINK,DISP=SHR
//SLSCNTL DD DSN=hlq.DBASEPRM,DISP=SHR
//SLSCNTL2 DD DSN=hlq.DBASESEC,DISP=SHR
//SLSSTBY DD DSN=hlq.DBASETBY,DISP=SHR
//SLSPRINT DD   SYSOUT=*
//SLSIN DD   *
CONFIG
 GLOBAL MAXVTV=32000 MVCFREE=40
 RECLAIM THRESHLD=70 MAXMVC=40  START=35
VTSS NAME=VSM401 LOW=70 HIGH=80 MAXMIG=8 RETAIN=5
 RTD  NAME=VSM42A00 DEVNO=2A00 CHANIF=0C
 RTD  NAME=VSM42A01 DEVNO=2A01 CHANIF=0D
 RTD  NAME=VSM42A02 DEVNO=2A02 CHANIF=0G
 RTD  NAME=VSM42A03 DEVNO=2A03 CHANIF=0H
 RTD  NAME=VSM42A04 DEVNO=2A04 CHANIF=0K
 RTD  NAME=VSM42A05 DEVNO=2A05 CHANIF=0L
 RTD  NAME=VSM42A06 DEVNO=2A06 CHANIF=0O
 RTD  NAME=VSM42A07 DEVNO=2A07 CHANIF=0P
 RTD  NAME=VSM42A08 DEVNO=2A08 CHANIF=1C
 RTD  NAME=VSM42A09 DEVNO=2A09 CHANIF=1D
 RTD  NAME=VSM42A0A DEVNO=2A0A CHANIF=1G
 RTD  NAME=VSM42A0B DEVNO=2A0B CHANIF=1H
 RTD  NAME=VSM42A0C DEVNO=2A0C CHANIF=1K
 RTD  NAME=VSM42A0D DEVNO=2A0D CHANIF=1L
 RTD  NAME=VSM42A0E DEVNO=2A0E CHANIF=1O
 RTD  NAME=VSM42A0F DEVNO=2A0F CHANIF=1P
 VTD LOW=9900 HIGH=99FF

VSM4 Configuration Example: 8 VCF Cards, 4 CLINKs, FICON Directors for 8 RTDs

Figure G-6 shows CONFIG channel interface identifiers for a VSM4 with 8 VCF cards. In this configuration, you have allocated:

8 Host ports.
4 ports for RTDs. The RTD ports are all connected to FICON directors, each of which is attached to RTDs, so the CHANIF identifiers for both RTDs are shown on each port. This allows Back-End connection to 8 RTDs. As with ESCON, only one RTD per port/Director can be active at a time.
4 ports for CLINK connections to form a Bi-Directional VTSS Cluster, and 8 ports to host connections. To form the clustered VTSS, you have two VSM4s (VSMPR1 and VSMPR2) configured identically as shown in Figure G-6. Bi-Directional Clustering requires pairs of Uni-Directional CLINKs with the FIPSs configured so that the data flows in opposite directions on the CLINKs. To make that happen, make 0G and 1G the sending (Nearlink Mode) ports on both VTSSs and 0O and 1O the receiving (Host Mode) on both VTSSs.

Figure G-6 VSM4 with 8 VCF Cards, 8 Host Ports, FICON Directors for 8 RTDs, 4 CLINK Ports

CONFIG Example for Bi-Directional Clustered VSM4 FICON Back-End

Figure G-7 shows an example of a Bi-Directional Cluster of two VSM4s (VSMPR1 and VSMPR2) with identical VCF card configurations shown in Figure G-6.

Caution:

Bi-Directional Clustering requires VTCS 6.1. You cannot configure a Bi-Directional Cluster at releases lower than VTCS 6.1.

Figure G-7 Dual ACS Bi-Directional Clustered VTSS Configuration

The follows example shows CONFIG JCL to define a Bi-Directional Cluster of two VSM4s (VSMPR1 and VSMPR2) as shown in Figure G-7. Note that:

The CLUSTER statement defines the Cluster as consisting of VSMPR1 and VSMPR2.

There are CLINK statements using the sending (Nearlink Mode) ports of both VTSSs to enable the Cluster as Bi-Directional. The Nearlink ports are 0G and 1G on both VTSSs.

//CREATECF EXEC PGM=SLUADMIN,PARM='MIXED'//STEPLIB DD DSN=hlq.SEALINK,DISP=SHR//SLSCNTL DD DSN=hlq.DBASEPRM,DISP=SHR//SLSCNTL2 DD DSN=hlq.DBASESEC,DISP=SHR//SLSSTBY DD DSN=hlq.DBASETBY,DISP=SHR//SLSPRINT DD   SYSOUT=*//SLSIN DD   * CONFIG RESET CDSLEVEL(V61ABOVE) GLOBAL MAXVTV=32000 MVCFREE=40  RECLAIM THRESHLD=70 MAXMVC=40  START=35
VTSS NAME=VSMPR1 LOW=70 HIGH=80 MAXMIG=8 MINMIG=4 RETAIN=5
 RTD  NAME=PR11A00 DEVNO=1A00 CHANIF=0C
 RTD  NAME=PR11A01 DEVNO=1A01 CHANIF=0D
 RTD  NAME=PR11A02 DEVNO=1A02 CHANIF=0K
 RTD  NAME=PR11A03 DEVNO=1A03 CHANIF=0L
 RTD  NAME=PR12A08 DEVNO=2A08 CHANIF=1C
 RTD  NAME=PR12A09 DEVNO=2A09 CHANIF=1D
 RTD  NAME=PR12A0A DEVNO=2A0A CHANIF=1K
 RTD  NAME=PR12A0B DEVNO=2A0B CHANIF=1L
 VTD LOW=9900 HIGH=99FF
 VTSS NAME=VSMPR2 LOW=70 HIGH=80 MAXMIG=8 MINMIG=4 RETAIN=5
 RTD  NAME=PR23A00 DEVNO=3A00 CHANIF=0C
 RTD  NAME=PR23A01 DEVNO=3A01 CHANIF=0D
 RTD  NAME=PR23A02 DEVNO=3A02 CHANIF=0K
 RTD  NAME=PR23A03 DEVNO=3A03 CHANIF=0L
 RTD  NAME=PR24A08 DEVNO=4A08 CHANIF=1C
 RTD  NAME=PR24A09 DEVNO=4A09 CHANIF=1D
 RTD  NAME=PR24A0A DEVNO=4A0A CHANIF=1K
 RTD  NAME=PR24A0B DEVNO=4A0B CHANIF=1L
 VTD LOW=9900 HIGH=99FF
 CLUSTER NAME=CLUSTER1 VTSSs(VSMPR1,VSMPR2)
 CLINK VTSS=VSMPR1 CHANIF=0G
 CLINK VTSS=VSMPR1 CHANIF=1G
 CLINK VTSS=VSMPR2 CHANIF=0G
 CLINK VTSS=VSMPR2 CHANIF=1G

IOCP Example for Single MVS Host Connected to a VSM4 Using FICON Directors

Figure G-8 shows a configuration diagram for a single MVS host connected to a VSM4 through FICON Directors, and the example that follows shows IOCP statements for this configuration. Note that:

From MVSA, you define 8 CHPIDs, with each path switched in the FICON Director, for a total of 8 channels running to the VSM4.
You code 16 CNTLUNIT statements to define the VSM4 as 16 3490 images.
You code IODEVICE statement to define the 16 VTDs that are associated with each 3490 image.

If ESCON and FICON channels are configured to the same logical control unit, MVS issues message CBDG489I, which indicates that mixing ESCON and FICON channel paths on a logical control unit should be used only for the migration from ESCON to native FICON but should not be used permanently. This is a warning message only, and does not indicate an error.

Figure G-8 Configuration Diagram: Single MVS Host Connected to a VSM4 Using FICON Directors

ESCD4C CHPID PATH=(20,70),TYPE=CNC,SWITCH=4C
ESCD4D CHPID PATH=(21,71),TYPE=CNC,SWITCH=4D
ESCD4E CHPID PATH=(30,80),TYPE=CNC,SWITCH=4E
ESCD4F CHPID PATH=(31,81),TYPE=CNC,SWITCH=4F

CU1 CNTLUNIT CUNUMBR=001,
       PATH=(20,21,30,31,70,71,80,81),
       LINK=(D0,D4,D0,D4,D4,D0,D4,D0),
       UNIT=3490,CUADD=0,
       UNITADD=((00,16))

STRING1  IODEVICE ADDRESS=(0500,16),
       CUNUMBER=(001),
       UNIT=3490,
       UNITADD=00,STADET=Y

CU2 CNTLUNIT CUNUMBR=002,
       PATH=(20,21,30,31,70,71,80,81),
       LINK=(D0,D4,D0,D4,D4,D0,D4,D0),
       UNIT=3490,CUADD=1,
       UNITADD=((00,16))

STRING2  IODEVICE ADDRESS=(0510,16),
       CUNUMBER=(002),
       UNIT=3490,
       UNITADD=00,STADET=Y
       .
       .
       .

CU15 CNTLUNIT CUNUMBR=015,
        PATH=(20,21,30,31,70,71,80,81),
        LINK=(D0,D4,D0,D4,D4,D0,D4,D0),
        UNIT=3490,CUADD=E,
        UNITADD=((00,16))

STRING15  IODEVICE ADDRESS=(05E0,16),
        CUNUMBER=(015),
        UNIT=3490,
        UNITADD=00,STADET=Y

CU16 CNTLUNIT CUNUMBR=016,
        PATH=(20,21,30,31,70,71,80,81),
        LINK=(D0,D4,D0,D4,D4,D0,D4,D0),
        UNIT=3490,CUADD=F,
        UNITADD=((00,16))

STRING16  IODEVICE ADDRESS=(05F0,16),
        CUNUMBER=(016),
        UNIT=3490,
        UNITADD=00,STADET=Y

Tip: Unlike ESCON, FICON supports multiple active I/Os per channel. If the number of active VTDs is less than the number of channels configured to the VTSS, the I/Os to those VTDs may not be evenly spread across all the channels. As the number of active VTDs increases to be greater than the number of channels configured to the VTSS, the channel subsystem will spread the I/Os across all the channels. If it is desired to spread the I/Os across all of the channels even when only a few VTDs are active, it is necessary to use the preferred path feature to force the channel subsystem to spread the I/Os across the channels. The preferred path feature is specified using the PATH= parameter on the IODEVICE statement. When you specify preferred path on the IODEVICE statement, the channel subsystem always tries the preferred path first. If it is busy or unavailable, the channel subsystem next tries the channel path following the preferred path in the rotation order, and so on.

The following example shows IODEVICE statements for STRING1 without using preferred pathing.

STRING1  IODEVICE ADDRESS=(0500,16),
       CUNUMBER=(001),
       UNIT=3490,
       UNITADD=00,STADET=Y

The following example shows IODEVICE statements for STRING1 using preferred pathing. If you are using preferred pathing, you need to use these kind of IODEVICE statements for all paths, such as STRING2 through STRING16 shown in the previous example.

STRING10 IODEVICE ADDRESS=(0500,2),
       CUNUMBER=(001),
       UNIT=3490,
       UNITADD=00,STADET=Y,
       PATH=20

STRING12 IODEVICE ADDRESS=(0502,2),
       CUNUMBER=(001),
       UNIT=3490,
       UNITADD=00,STADET=Y,
       PATH=21

STRING14 IODEVICE ADDRESS=(0504,2),
       CUNUMBER=(001),
       UNIT=3490,
       UNITADD=00,STADET=Y,
       PATH=30

STRING16 IODEVICE ADDRESS=(0506,2),
       CUNUMBER=(001),
       UNIT=3490,
       UNITADD=00,STADET=Y,
       PATH=31

STRING18 IODEVICE ADDRESS=(0508,2),
       CUNUMBER=(001),
       UNIT=3490,
       UNITADD=00,STADET=Y,
       PATH=70

STRING1A IODEVICE ADDRESS=(050A,2),
       CUNUMBER=(001),
       UNIT=3490,
       UNITADD=00,STADET=Y,
       PATH=71

STRING1C IODEVICE ADDRESS=(050C,2),
       CUNUMBER=(001),
       UNIT=3490,
       UNITADD=00,STADET=Y,
       PATH=80

STRING1E IODEVICE ADDRESS=(050E,2),
       CUNUMBER=(001),
       UNIT=3490,
       UNITADD=00,STADET=Y,
       PATH=81