IBM Student Mainframe Challenge Part Two · IBM Student Mainframe Challenge Part Two Time to complete – about six to eight hours ... people use it as a verb: “to IPL” meaning

IBM Student Mainframe Challenge

Part TwoTime to complete – about six to eight hours

The company that you work for acts as a broker for various suppliers, and provides a

website that customers can use to purchase products that they sell. In this part of the

Mainframe Challenge, we will use CICS Transaction Server to update order information,

create a batch job to extract sales data from a DB2 database and then generate a sales

report from it. Before we do this though, let's have a brief look at some system settings to

introduce you to how a z/OS system can be configured by its administrators.

Help

You may find the following references useful when completing the tasks:

– z/OS v1.11 Information Center:

http://publib.boulder.ibm.com/infocenter/zos/v1r11/index.jsp

– CICS Transaction Server v3.2 Information Center:

http://publib.boulder.ibm.com/infocenter/cicsts/v3r2/index.jsp

GETTING STARTED

Before you can start this part of the contest you'll need to obtain the questions that you need to

answer along the way.

These are available in a question and answer sheet in the sequential data set called

ZOS.CONTEST1.PART2.ANSWER.SHEET.

➔ Using ISPF take a copy of this data set, which you can complete as you progress through

the tasks.

Call your sequential data set UKxxxxx.UKxxxxx.ANSWERS

It should be allocated in tracks (TRKS)

It should have a primary quantity of 1 and secondary quantity of 1

It should have 0 directory blocks

The record format should be FB (fixed block) and the record length should be 80

The block size should be 32000

The data set name type should be blank

➔ Ensure that the first TWO qualifiers of your data set are your user id otherwise your

answers may not be marked correctly.

To copy a data set, use utility 3.3 from the ISPF primary panel. (Type =3.3 at

the command line in ISPF.)

You can also copy PDSE members fromthis panel, by putting the member name

in brackets after the data set name.

To test out your new z/OS skills, I'd like you towork with the order processing team, but

first Gemma will hand you over to Carlos, one of our system programmers. To check your progress and understanding, I have some

questions for you to answer along the way.

You will be advised when you should be able to answer each of the questions, but please read the

comments at the top of the answer sheet carefully before proceeding.

➔ Fill in your z/OS user id (i.e. UKxxxxx) in the space provided above question 1.

If it all goes wrong, don't worry! Delete thedataset by finding it in utility 3.4, and

typing D in the command column on the left.Then try again!

INTRODUCTION TO SYSTEM PROGRAMMING

Mainframe systems are designed to be used by large numbers of people. Most of those who interact

with mainframes are end users - people who use the applications that are hosted on the system.

However, because of the complexity of the mainframe software that supports the large number of users

and applications, a variety of roles are needed to operate the mainframe itself.

z/OS provides many commands and facilities to help the System Programmer achieve this and we will

now look at a few of these.

When answering the following questions you might find the MVS book within the z/OS Information

Center useful to understand a command's syntax, what it is used for, and the output that it produces.

You can find the Information Center at:

http://publib.boulder.ibm.com/infocenter/zos/v1r11/index.jsp

One of those roles is System Programmer -that's me! I'm Carlos; I'm responsible for theinstallation, customisation and maintenance

of z/OS on our mainframe.

USING SDSF

In this section, you may find yourself having to repeatedly swap between two different applications, for

example, between the answer sheet data set in the editor, and SDSF to find out the answers to the

questions. It can get frustrating to keep having to go in and out of different applications and back

again, so there is a better way to do it:

Pressing F2 will cause the screen to "split" at your cursor, showing the two screens

separated by a horizontal dotted line. You can then press F9 to swap between the two

screens.

To maximise the available screen size, put your cursor on the very top line of the screen

before pressing F2.

SDSF Primer

System Display and Search Facility (SDSF) provides information that you can use to monitor,

manage and control a z/OS system.

You can use SDSF to view the system log and the status of jobs that are running in the system or have finished running, including any held output of jobs that have not yet been purged (discarded).

You access SDSF from the ISPF primary panel with the command S.

From there, DA can be used to view the active (running) jobs in the system, including any

started tasks, which are a special type of job.

ST can be used to view the status of jobs, those that are active, those yet to be processed,

and those that have completed.

You'll never know howyou lived without it!

Once you are viewing a list of jobs, at the command input prompt you can filter the list of

jobs to just those that you own by entering:OWNER UKxxxxx

To revert back to showing all jobs enter:OWNER *

Similarly you can filter the list of jobs by their name by entering:PREFIX ABC*

which would show only those jobs whose names begin with the letters ABC.

To remove this filter enter:PREFIX *

You can use both the owner and the prefix filters at the same time.

These are particularly useful because they hide jobs on the system that you are either not

interested in or not authorized to look at.

Be aware that SDSF remembers the filters that you are using. If you are not seeing a job

that you expect to see, set both filters to a single asterisk (*) so that all jobs are displayed.

In the DA and ST panels you are presented with a list of jobs. Use the PF7 and PF8 keys to

page up and down.

To view more information about a particular job you can enter S in the left-hand column

(NP) and press Ctrl.

This can provide a lot of data and so alternatively you can enter a question mark (?) instead,

which will give a listing of the various sections of the output. You can then use S to select

these sections individually.

Use the PF3 key to exit the screen that you are on and return to the previous one.

The prefix and owner filters aren't displayedanywhere by default, so if you leave the filterson you might end up with a blank job list! ISPFremembers your filters between sessions, solook in the Options menu at the top of thescreen and choose option 5 to display them.

INTRODUCTION TO IPL

When you turn on your desktop or laptop at home the operating system starts via a process known as

“booting”. This process uses a boot record on disk to locate the operating system and initiate the

startup process. On z/OS this is known as “IPLing”, and the equivalent of a boot record is known as a

load member.

From the z/OS Primary Option Menu in ISPF select the option for SDSF and then the ULOG option that

opens the User Session Log, which can be used to look at just the output for commands that you issue.

This saves having to scroll up and down the main system log, which can have lots of entries, to locate

the appropriate lines.

To look at the IPL information for the z/OS system issue the /DISPLAY IPLINFO command, or if you

prefer, /D IPLINFO instead. The response will be output to your session log, take a moment to have a

read of it.

➔ Now answer question 1.

IPL stands for “Initial Program Load”, butpeople use it as a verb: “to IPL” meaning

“to start the mainframe”. It's a good place tostart learning about system programming, sowe'll have a look at the info provided to our

z/OS system when it was IPL'd.

In addition to the load member, the system also requires one or more data sets that are collectively

known as the System Parameter Library (PARMLIB). These data sets contain various members that

detail additional configuration information for the different aspects of the system.

➔ Now answer question 2.

Display commands are listed under MVS – System Commands in the z/OS Infocenter

If you're more familiar with Linux or UNIX,the PARMLIB on z/OS is a bit like the

/etc/ directory.

The PARMLIB configuration settings on z/OS are divided into different members for different aspects of

the system. z/OS builds the names of these members using specific prefixes combined with

customizable suffixes. Most of these suffixes are defined in the system parameter list, which can be

found in one or more members whose names start with IEASYS.

The use of suffixes such as this allows an operator to IPL a z/OS system with different settings simply

by specifying different suffixes for the IEASYS members. The default IEASYS suffixes are generally

specified in the load member but an operator may override this when an IPL is started.

You should see that this list contains one entry for the suffix 00; the (OP) afterwards, if present,

indicates that it was specified by the operator, not by the load member. From this we can determine

that when the system was IPLed it read a member called IEASYS00.

➔ Locate and browse this member in the PARMLIB data set.

You should be able to see how the suffixes for the other configuration settings are defined.

➔ Now answer question 3.

The DISPLAY IPLINFO command you issueda minute ago displayed the list of suffixesthis system used. Look at the top of the

ULOG for the output –you should see a line for IEASYS LIST

System resources

When going to buy a personal computer most people like to look at the specifications, such as how

much memory it has. z/OS refers to memory (RAM) as real or central storage. The physical System z

machine can run a large number of z/OS systems, known as logical partitions (LPARs). Each LPAR has

a subset of the physical resources allocated to it, either exclusively or shared with other LPARs.

To display the amount of real storage that has been made available to the LPAR for the z/OS system

you are using you can use the DISPLAY M=STOR command.

➔ Issue this command and then answer question 4.

z/OS also provides a UNIX environment, known as UNIX System Services (USS) or OMVS (it used to be

called OpenEdition MVS). You will get a chance to use USS in part 3 of the contest.

Just like UNIX systems you can configure many options, such as how many processes can run

concurrently, which is not related to the number of concurrent jobs that you identified earlier. These

can be found in one of the PARMLIB members, but it is much simpler to issue the DISPLAY OMVS,L

command to view the configured limits.

➔ Issue this command and then answer question 5.

BATCH JOBS

In this part of the contest you will be using batch jobs to perform work. To run a batch job you have to

submit it and it then gets queued up to be executed until an initiator (a process that runs jobs)

becomes free.

To help the system balance and prioritise the various jobs that might be submitted each job is assigned

a classification (class) and each initiator is configured to run jobs of one or more classes. This allows

the System Programmer to easily configure the system to allow, for example, twice as many jobs of

one class to be run at the same time as that of another, less important classification.

If a job is submitted for a class for which there is no initiator it remains in the queue indefinitely until it

is either purged or an initiator for it is made available.

➔ Exit out of the user session log and select this option from the SDSF Primary Option Menu.

From this screen you can see the initiators, their status and what job classes they have been configured

to run. An active initiator is currently running a job, an inactive one is currently idle (waiting for work),

and a drained initiator is currently disabled.

➔ Now answer question 6.

You can see the various initiators that havebeen configured on our z/OS system using

another SDSF option. See if you can guess which one!

There's plenty more to system programmingthan what you've seen here, but I think

that's enough for now. I'll hand you back to Gemma, and you can run some jobs of your

own! Don't break anything!

PART 2 - TASKS

Information about the products that are sold by each retailer and the orders that customers have

placed are stored in a DB2 database. The order processing team uses a CICS transaction which

connects to DB2 to extract and update the necessary data.

Welcome to the order processing team!

...or, as I like to call it, theawesome processing team!

Our company's strategy is to provide a websitethat allows customers to place orders for

products sold by several other retailers. Theorder processing team can review recent ordersand change their state as they are processed.

INTRODUCTION TO CICS

CICS (Customer Information Control System) Transaction Server is IBM's general-purpose transaction

processing software for z/OS. It is a powerful application server that meets the transaction-processing

needs of both large and small enterprises. It builds on z/OS and System z facilities to provide high

availability and scalability at a low cost per transaction; it supports large transaction volumes with a fast

and consistent response time.

Transaction processing is a style of computing, typically performed by large server computers, that

supports interactive applications. In transaction processing, work is divided into individual, indivisible

operations, called transactions. For example, withdrawing cash from an ATM and updating the balance

in a database. By contrast, batch processing is a style of computing in which one or more programs

processes a series of records (a batch) with little or no action from the user or operator. For example,

printing out bank statements for every account holder and sending them in the mail.

Transaction processing middleware such as CICS allows application programmers to concentrate on

writing code that supports the business, by shielding application programs from the details of

transaction management:

• It manages the concurrent processing of transactions.

• It enables the sharing of data.

• It ensures the integrity of data

• It manages the prioritization of transaction execution.

When a transaction starts processing, CICS runs a program that is associated with the transaction. That

program can transfer control to other programs in the course of the transaction, making it possible to

assemble modular applications consisting of many CICS programs.

At any time, in a CICS system, many instances of a transaction can run at the same time. It would be

terrible to have to wait for other people to finish paying before you could buy something online.

CICS Primer

Now's a good time to give you some guidance on using CICS.

You start a CICS transaction by pressing the CLEAR* key to clear the screen, and entering

the transaction identifier, either by itself or followed by data, on the command line of the

screen. The command line is a single line, usually at the top of the screen.

You can type the transaction identifier by itself and follow the prompts until a complete

transaction command is built up, or you can type the complete transaction command on the

command line. If you do not enter enough information, or if the information you enter is

wrong, you are prompted to complete or correct your input.

For example, in the following transaction, CESF is the transaction identifier and the

additional data is GOODNIGHT.

CESF GOODNIGHT

This transaction is used to log off CICS.

When the transaction starts, it processes the additional data. At the completion of this

transaction, you get the following message:

STATUS: SESSION ENDED

After a transaction has completed, press the CLEAR* key to get a blank screen in readiness

for the next transaction.

To end a transaction use the PF3 key.

* The special 3270 CLEAR key is not on most keyboards. However, 3270 emulators either

provide this via a context (right-click) menu or allow you to map it to a key of your choice,

often Pause/Break by default.

You can find more information on CICS in the CICS Transaction Server v3.2 Information Center.

Using CICS

Open a second 3270 session but instead of entering TSO at the SELECT APPLICATION prompt enter

LOGON APPLID(CICSZ057) instead.

You should then be presented with the CICS sign-on screen:

➔ Enter your user id and password and press enter (remember this is the Ctrl key).

The screen should clear except for the following message near the bottom:

DFHCE3549 Sign-on is complete (Language ENU).

Viewing and updating orders in CICS

➔ Once you have successfully logged on to CICS, run a transaction called ORDS.

Item TN032 from Camping Stuff has been on order for S Carter for quite a long time and she has

now decided to give up and cancel the order.

➔ Change the status of this order from On Order to Cancelled.

The order for item GF157 from Sally's Sports by J Bloggs has been accidentally cancelled by one

of your colleagues.

➔ Change the status of this item from Cancelled to Packing.

➔ Now answer questions 7 to 9.

This is the transaction we use to displayrecent customer orders. We'll changethe state of a couple of orders, then I

have some questions for you to answer!

INTRODUCTION TO JCL

Job Control Language (JCL) is how you describe this information to z/OS, whereby a job represents a

task.

For every job that you submit, you need to tell z/OS where to find the appropriate input, how to

process that input (that is, what program or programs to run), and what to do with the resulting

output. You use JCL to convey this information to z/OS through a set of statements known as job

control statements.

JCL's set of job control statements is quite large, enabling you to provide a great deal of information to

z/OS. Most jobs, however, can be run using a very small subset of these control statements.

Within each job, the control statements are grouped into job steps. A job step consists of all the control

statements needed to run one program.

If a job needs to run more than one program, the job would contain a different job step for each of

those programs.

Job Control Statements

Every job must contain:

- A JOB statement to mark the beginning of a job and assign it a name. This statement is

also used to provide certain administrative information, including security, accounting, and

identification information. Every job has one and only one JOB statement.

- At least one EXEC (execute) statement, marking the beginning of a job step, a name for

the job step, and the program or procedure to execute. You can add various parameters to

the EXEC statement to customize the way the program executes and the conditions under

which the step should be run.

How would you achieve a task that wasgiven to you? I'd break it down into oneor more steps, each completing a part

of the bigger picture. Sometimes I'll needdifferent tools and resources for each step.

Most programs require some input or generate some output and so usually also contain:

- One or more DD (data definition) statements, to identify and describe the input and

output data to be used in the step. This statement may be used to request a previously

created data set, to define a new data set, to define a temporary data set, or to define and

specify the characteristics of the output.

Each job control statement has five fields:

1. An identifier field, which is typically two forward slashes (//). The entire line is treated as

a comment if the two slashes are followed by an asterisk (//*).

2. A name field, identifying the statement so that it can be referred to later

3. An operation field identifying the type of statement, i.e. JOB, EXEC, DD.

4. A parameter field

5. Anything following the parameter field is treated as a comment and ignored.

JCL statements can be coded only up to column 71. You may question why JCL has this format but it

was historically input using punch cards that had 80 columns, the last 8 of which were used for line

numbers.

For more information on JCL you can look in the JCL Reference section of the MVS book in the z/OS

Information Center.

JCL's layout can look a bit confusing whenyou see it for the first time. Luckily the

ISPF editor can display JCL syntaxhighlighting if you type HILITE JCL

(or HILITE AUTO) when editing(not browsing) a JCL member.

GETTING STARTED

➔ Use ISPF to create two new PDSEs called UKxxxxx.PARTTWO.JCL and UKxxxxx.PARTTWO.REXX

They should be allocated in tracks (TRKS)

They should have a primary quantity of 1 and secondary quantity of 1

The record format should be FB (fixed block) and the record length should be 80

The block size should be 32000

The data set name type should be LIBRARY

Extract the order data from DB2

The existing program to extract all sales data since the beginning of the year from the DB2 database is

called DB2EXTR and is stored in ZOS.CONTEST1.PART2.LOAD. In order to execute this program you will

need to submit a JCL job.

We did something similar to this last year......aha! We used an existing application toextract sales data from the database, thenran another program to produce the report

from that extracted data. They're bothrun within batch jobs using JCL.Let's get them working again...

For Part 2 we're going to be using both JCLand REXX, so we can create datasets tostore both of them now. We won't put

anything in the REXX dataset until later.

If you've forgotten how, go back to Part 1!

The head of Sales and Marketing is asking me fora report of the total sales to date this year for

each retailer. She's also interested in whichitems the retailers are selling most and least of,so that marketing campaigns can be targeted.

ZOS.CONTEST1.PART2.JCL(RUNDB2AP) is an example of a job to run a DB2 application, so it's a good

starting point for your job.

➔ Copy this into your JCL library, and rename the member to DB2EXTR.

You can rename the new copied member either by giving it a different name in the copy dataset

ISPF panel, or by browsing the dataset it has been copied to and putting r next to the member.

If you choose the second option, you'll have to type REFRESH on the command line to update the

list with the member's new name.

There isn't any JCL to run the first program,so you'll need to write some yourself.

Don't worry, hardly anyone writes JCL fromscratch these days – it's much easier to

modify a job that already exists!

or

The first line in the job contains a JOB statement that assigns a name to the job.

➔ Change the name of the job to DB2EXTR.

The next statement in the job is an EXEC statement, which marks the beginning of a step in the job

and causes a program to be executed. In this case, the program being executed is IKJEFT01, the TSO

Terminal Monitor Program. This program will execute the commands specified in the SYSTSIN data

definition statement.

The DSN command connects the application to a particular DB2 subsystem.

➔ Change the name of the subsystem to DB1S.

The RUN command runs the application program.

➔ Change both the name of the program and the DB2 plan name to DB2EXTR, and the

library where the program can be found to ZOS.CONTEST1.PART2.LOAD.

➔ Submit the DB2EXTR job and check the output.

It turns out that this job contains a parameter that requests that the system validates the syntax of the

JCL you have written without proceeding to execute the job. This parameter can be useful for checking

for errors when writing or modifying jobs, but you will need to remove it and submit the job once again

in order to run the program.

➔ Check the job output in SDSF to see if it completed successfully.

There doesn't seem to be any output fromthe program execution, and it executedfor zero time. I don't think the program

was executed by the job...

Hmm, more problems. Your job isn'tcomplete yet!

The DB2EXTR program outputs the data it extracts from the database to a file called ORDERDAT.

However, this isn't the name of the data set because z/OS uses symbolic file names to provide naming

redirection between the file name used within the program and the actual data set used during the

execution of the program. This indirection allows the same program to be run against different data

without needing to change it or add support for passing in many parameters.

The JCL data definition (DD) statement allows you to define a file in a job step that will be available to

the program being executed, and specify the data set that this file refers to.

Your job was unable to open the output file as there was no corresponding DD statement in the job

step. Therefore, you need to add a DD statement named to allocate a data set that the program can

use for output.

➔ Call new new DD statement ORDERDAT, and specify the necessary parameters so that

the data set will be allocated with the following attributes:

Use the DSN parameter to specify a data set name of UKxxxxx.HIST.ORDS

Use the SPACE parameter to specify a primary and secondary space allocation of 1 cylinder.

Use the DISP parameter to allocate a new data set when the job step is executed and

catalog the data set when the step has completed successfully

➔ While you're doing this, answer questions 10 and 11.

Now would be a good time to look in thez/OS Infocenter if you haven't already, under

JCL reference in the MVS section.

Here's a hint, the DISP parameter will needtwo subparameters!

A brief explanation of storage management

When you look at the details of the DISP parameter of the data definition statement in the

MVS JCL Reference book, you will see references to SMS-managed data sets. Data sets on

z/OS are allocated on volumes, in the same way that files on your PC are allocated on your

hard disk. However, a z/OS system will have tens, or even hundreds of volumes available.

This makes it difficult for users to know which volume to use when allocating new data sets,

and for storage administrators to balance free space across volumes. SMS (System Managed

Storage) solves this problem by enabling the system to choose which volume a new data set

should be allocated on using various criteria set by the storage administrator. Data sets

allocated in this manner are called SMS-managed data sets. Most data sets on modern

systems, including any that you allocate, will be SMS-managed.

You will also see references to cataloging data sets. The catalog is the z/OS component that

allows you to locate which volume a data set with a particular name resides on. This allows

you to refer to the data set in a job using only its name. It is possible to allocate a data set

without adding it to the catalog, but in order to reference an uncataloged data set you

would need to specify both the data set name and the volume where it resides.

➔ Now that you've made those changes, submit the DB2EXTR job again.

It should with a return code of zero, which indicates that it completed successfully. Browse the output

data set to confirm that the program has successfully extracted the sales data from the database, and

to familiarise yourself with the format of the data.

Note: If you run the job more than once, you will need to delete the data set it creates

first. If you do not do this, you will get a JCL Error and a message in the job output

indicating that the new data set couldn't be allocated due to a duplicate data set name.

INTRODUCTION TO REXX

REXX (which stands for REstructured eXtended eXecutor) is an interpreted programming language that

was designed by IBM to be easy to read and learn while also providing powerful capabilities. As it is an

interpreted language there is no need to compile the program before it is executed – the REXX

command processor directly processes each language statement as it is encountered during the

execution of the program. However, there is a REXX compiler available should it be desirable to improve

the program's performance or source code security.

For more information on the structure and syntax of the REXX language, and descriptions of the

operators and built-in functions provided, refer to the TSO/E REXX User's Guide and the TSO/E REXX

Reference sections of the z/OS Information Center.

Some useful REXX tips:

• Issue the HILITE REXX command in the ISPF editor to get syntax highlighting when

editing REXX code.

• A comma at the end of a line indicates that the current line of REXX code is continued on

the next line.

• Variables of the format x.y in REXX are called compound variables, and are similar to

arrays. Check the REXX User's Guide and REXX Reference for details of how to use these

variables.

Produce the report from the extracted data

REXX is a language that is very well suited to the task of producing reports like these due to its

character manipulation capabilities.

REXX has been around for a while, butit's still widely used on the mainframe.

I like it because it's pretty laid-back aboutsyntax rules – variables are case-insensitive,don't have fixed datatypes and don't have to

be defined before they're used! It's alsoreally handy for parsing text data, thereare plenty of built in functions for that.

Now you've extracted the sales data fromthe database, you need to run another

program to generate the report. I can giveyou the program and the JCL job this time!

➔ Copy the job from ZOS.CONTEST1.PART2.JCL(GENREPRT) to your JCL library, and the

program source from ZOS.CONTEST1.PART2.REXX(GENREPRT) to your REXX library.

You'll notice that this job has two steps. The first step deletes the report data set if it exists already,

which is very useful if you need to submit the job more than once!

➔ Replace the ++XXXX++ placeholders appropriately in the job, calling your report data

set UKxxxxx.SALES.REPORT and submit it.

➔ Correct the error, and submit the job again.

➔ Correct the error in the program, and any others you may encounter.

This time it should execute without any errors, and a sales report should be generated. However, it

doesn't look as though the report of the best and worst selling items has been sorted correctly.

➔ Correct this problem in the REXX code, and resubmit the job.

I said I'd give you the JCL job, I never saidit would work first time!

It looks like the REXX program didn't workfirst time either. We didn't pick up on this

earlier because REXX is interpreted – it hadn'treached this point when we found the first

error. If we'd compiled it we'd have found out.

Your report says that the best selling itemshave fewer sales than the worst selling ones!This report is sorted backwards! Please fix it.

Improve your JCL

Instead of having to submit two separate jobs to extract the sales data from DB2 and generate the

report, this can all be done in one job that has two steps (three steps if you include the first step that

deletes the current report data set).

➔ Change your GENREPRT job so that it performs both tasks.

You might find it helpful to also modify the first step of the job so that it deletes both the old report and

extracted sales data sets.

A special type of data set called a temporary data set can be used for this purpose. Temporary data

sets are created and deleted within a single job.

➔ Change the data set used to store the sales data extracted from the database to be a temporary data set.

➔ While you're doing this, answer questions 12 and 13.

Hint: You will need to change the name of the data set to follow a certain format as this is

what tells the system that the data set is temporary, and also change the data set

disposition such that the data set allocated in the first step is passed on for use in the

second step.

Great job! You fixed up the JCL and the REXX,and managed to generate the report the bosswanted! But I reckon you can improve the JCLeven further – can you combine this job and

the DB2EXTR job we used earlier?

The dataset which has sales data extractedfrom the database is passed to the secondstep which generates the report. It doesn'tneed to be kept after the job has finished.

Submit your answers

➔ Review your answers and then submit them so that they can be marked. Your report will

also be checked but you do not have to submit it in the same way. Do however ensure that it contains data for the whole year and not just for a single month.

To do this go to option 6 (Command - Enter TSO or Workstation commands) and enter:

XMIT ZOS19.MATTK DA('UKxxxxx.UKxxxxx.ANSWERS')

You've achieved a lot so far, but there's stillplenty more to do! Part Three builds on what

you've been learning, and gets even morechallenging! You'll be adding functionality tothe customer website, and using another of

IBM's world-famous products!

Congratulations! You've finished all the steps forPart Two! Now we can target our advertisingtowards the most popular products that wesell. If you're one of the first to complete

the task and answer the questions correctly,I'll send you a prize in due course!

I'm recommending you to our events teamto carry on to Part Three of the contest –

you'll need access to their mainframe systems.Please send an email to your contest task ID to request a new user ID, including in your

email your current user ID. Good luck!