2009 WebSphere Services Technical Conference

sweetleafapartInternet and Web Development

Aug 7, 2012 (5 years and 5 months ago)

626 views


© 2009 IBM Corporation



2009 WebSphere Services Technical Conference

world class skill building and technical enablement


February 9
-

13, 2009



Las Vegas, NV

WebSphere on z/OS


Like Thunder and Lightning

it Ain’t so Frightening!

Session Number: X01

Edward McCarthy

edwardmc@au1.ibm.com

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

2

Objectives


WebSphere on z/OS and distributed


What is the same


What is different



The BIG difference



Clustering



WebSphere on z/OS V7


New and unique functionality



WAS on z/OS


the good life



Production on one box



2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

3

Setting the scene…


You are a WebSphere Administrator


Who has only worked on Windows and/or Unix



Then one day


The sky darkens…


The thunder and lightening starts…


The phone rings, it’s your boss, with news…



We’re moving to WebSphere on z/OS

After this call,

You are probably here

on this curve

After this session,

hope to have you closer to

here

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

4

Similarities


Code base



Since V6.0, code base for WebSphere on z/OS same as used on
distributed


Which is since Mar 2005


Includes


Portal


Process Server


etc


Has extra code to take advantage of z/OS



Things that are the same:



J2EE Specification support


Terminology


Product and maintenance release dates


Administration



2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

5

J2EE Specification support


J2EE Applications


Written to the specification


Will run unchanged in WAS on z/OS


No need to recompile


IBM Techdoc: Moving Applications to WebSphere on
z/OS


http://www
-
03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/WP101093


2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

6

Terminology


WebSphere Application Server


Is the name of the product


Also used to refer to the actual process that runs the application code


Server


This is the component that has the Java Virtual Machine ( JVM )


This is where the application programs run


Cluster


Logical term used to describe a group of servers


Node


Logical term to describe a single machine that runs one or more
servers


Cell


Logical term that covers the complete WebSphere configuration



Daemon


Separate process required on z/OS


Small component, needs little attention

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

7

Maintenance levels


WebSphere on z/OS


Uses same maintenance nomenclature



From log on z/OS


BBOM0007I CURRENT CB SERVICE LEVEL IS
build level
6.1.0.12

(
cf120738.13
) release
WAS61.ZNATV date 09/25/07 00:03:32.



From log on Windows


WebSphere Platform 6.1 [BASE
6.1.0.19
cf190836.04
]





2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

8

z/OS and Distributed


which Admin GUI is
which?

z/OS

Distributed

In other words


same look and feel

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

9

wsadmin on z/OS and distributed

WZADMIN @ SC55:/WebSphereEd/wzcell/dmgr/DeploymentManager/profiles/default/bin>./wsadmin.sh

port
7010

user wzadmin

password xyz

lang jython

WASX7209I: Connected to process "dmgr" on node wzdmnode using SOAP connector; The type of process
is: DeploymentManager

WASX7031I: For help, enter: "print Help.help()"

wsadmin>AdminControl.completeObjectName("type=DeploymentManager,*")

'WebSphere:name=DeploymentManager,process=dmgr,platform=common,node=wzdmnode,diagnosticProvid
er=true,version=6.1.0.12,type=DeploymentManager,mbeanIdentifier=DeploymentManager,cell=wzcell,spec=1
.0'

C:
\
zProducts
\
was61
\
AppServer
\
profiles
\
Dmgr01
\
bin>wsadmin
-
lang jython

WASX7209I: Connected to process "dmgr" on node Dmgr01 using SOAP connector; The type of process
is: DeploymentManager

WASX7031I: For help, enter: "print Help.help()"

wsadmin>AdminControl.completeObjectName("type=DeploymentManager,*")

'WebSphere:name=DeploymentManager,process=dmgr,platform=common,node=Dmgr01,diagnosticProvide
r=true,version=6.1.0.9,type=DeploymentManager,mbeanIdentifier=DeploymentManager,cell=Dmgr01,spec=
1.0
'

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

10

Tracing via Admin GUI


z/OS and Distributed

z/OS

Distributed

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

11

Trace via wsadmin on z/OS and distributed

WASX7209I: Connected to process "dmgr" on node wzdmnode using SOAP connector; The type of process
is: DeploymentManager

WASX7031I: For help, enter: "print Help.help()"

wsadmin>ts = AdminControl.completeObjectName('type=TraceService,process=wzsr01a,*')

wsadmin>AdminControl.setAttribute(ts, 'traceSpecification', 'com.ibm.*=all=enabled')

WASX7209I: Connected to process "dmgr" on node Dmgr01 using SOAP connector; The type of process is:
DeploymentManager

WASX7031I: For help, enter: "print Help.help()"

wsadmin>ts = AdminControl.completeObjectName('type=TraceService,process=server
-
1,*')

wsadmin>AdminControl.setAttribute(ts, 'traceSpecification', 'com.ibm.*=all=enabled')'

BossLog: { 0096} 2008/09/25 07:10:48.934 01 SYSTEM=SC55 SERVER=WZSR01A PID=0X02010237

./bborjtr.cpp+440412145 ... BBOO0222I: TRAS0018I: The trace state has changed. The new trace state is

*=info:com.ibm.*=all.

[25/09/08 17:17:05:099 EST] 0000002c ManagerAdmin I

TRAS0018I: The trace state has changed. The new trace state is *=info:com.ibm.*=all.

z/OS

Distributed

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

12

In short…


From an administration point of view


Lots of similarities


Admin Gui


same


wsadmin


same


Terminology
-

same



If you have skills as a administrator for
WebSphere on Windows or Unix


Then those same skills transfer seamlessly to
WebSphere on z/OS


2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

13

So what is different ?


Working on z/OS


Information about running WAS processes


Logs


Threads in the JVM


Installation


Configuration

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

14

Working on z/OS


Windows


Has its user interface


the GUI we’ve all grown to love ;
-
)


Unix


Has its user interface


Either command line or GUI


z/OS


Has its own user interface as well


Several in fact


TSO, ISPF


Telnet


Rational Application Developer for z



Bottom line


The WebSphere administrator for WebSphere on z/OS


needs to know some TSO, ISPF basics


But does not need to be a z/OS guru

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

15

ISPF


command interface

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

16

Telnet into z/OS

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

17

Rational Application Developer for z


Eclipse based


User friendly way to
work on z/OS

Work with Unix files

Work with datasets

View job output

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

18

What WebSphere processes are running?

Windows

Task Manager

Unix

top

Q: Which java process is the DMGR and which is the node agent?

Q: Can you monitor WebSphere processes on other machines?

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

19

On z/OS
-

the view from SDSF

JOBNAME StepName ProcStep JobID Owner Real


WZDEMN WZDEMN BBODAEMN STC07908 WZDMN 6785

WZDMGR WZDMGR BBOCTL STC07906 WZDMCR1 70T

WZDMGRS WZDMGRS BBOSR STC07910 WZDMSR1 101T

WZNODE5 WZNODE5 BBOCTL STC07909 WZASCR1 43T

WZSR01A WZSR01A BBOCTL STC07923 WZASCR1 67T

WZSR01AS WZSR01AS BBOSR STC08092 WZASSR1 86T

WZSR01AS WZSR01AS BBOSR STC08065 WZSRSR1 87T

WZSR01AA WZSR01AS BBOSR STC08066 WZSRSR1 85T

WZSR64A WZSR64A BBGCTL STC07928 WZASCR1 88T

WZSR64AS WZSR64AS BBGSR STC08091 WZASSR1 255T

Deployment

Manager

Node Agent

Server

Daemon

Server

Control

Servants

Adjunct

Via SDSF

-

with a good naming convention, able to determine which STC is DMGR etc

-

Able to view WebSphere STCs running on any LPAR in the Sysplex

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

20

WebSphere logs


On distributed


SystemOut.log


SystemErr.log


native_stderr.log


Verbose Garbage collection



On z/OS



What is normally written to these files is written to
z/OS spool

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

21

WebSphere logs on z/OS

SystemErr.log

native_stderr.log

SystemOut.log


This is default setup


Custom properties can be used to write old log data from spool to a file


ras_stderr_ff_interval, ras_stdout_ff_interval


Can change JCL so that SystemOut and SystemErr


Are written to files


But no rolling capability


See:


http://www
-
03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/TD101087

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

22

Threads in the JVM


On Distributed


Can set number of threads to any value



On z/OS


Number of threads determined by workload profile selected



Workload Profile


Set via wsadmin or adminconsole:


Servers >> Application Servers >> server_name >> ORB Service>> Advanced Settings >>
"Workload Profile“


Can be set to:


ISOLATE (1 thread)


NORMAL (3 threads)


CPUBOUND (# of CPs
-
1, minimum of 3)


IOBOUND (Number of CPs*3, Min=5, Max=30)


LONGWAIT (40)


V7


provides property to set custom value



WebSphere for z/OS doesn't need threads as placeholders for work


WLM queues are used for that



Plan for # of in and ready threads to be 2
-
3X the # of CPs



Experiment with these values and the number of servant regions to optimize your
performance.


Too many servant regions take excessive storage


Too many threads in a JVM creates interference & more frequent GC.

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

23

Installation


On Windows and Unix


Typically run WebSphere supplied install GUI or install script


Which installs the software into some specified location



On z/OS


All z/OS software installed via z/OS mechanism called SMP/E


SMP/E has been in use for over 20 years for software install


Typically done by your friendly neighbourhood z/OS System programmer



Can have different versions of WebSphere installed at the same time


In fact can have different maintenance levels of a WebSphere version
installed at the same time


And in use

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

24

Configuration on Distributed


On Windows and Unix


Can use GUI


Or run WebSphere supplied command:


manageprofiles


To create profiles for nodes etc



On z/OS


Different process


If you’re going to get anywhere with this then you..


Need to accept it is a different process


Be willing to learn



If you are new to z/OS


You need someone with z/OS experience to assist you


2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

25

Configuration on z/OS


Process to build a cell:



Use TSO/ISPF or Windows Eclipse based tool


In which supply various values such as:


Started task names


Security related userid’s


TCPIP port numbers


Cell, node and server names


Generates small number of batch jobs


Run batch jobs to create cell



Keep sense of perspective



Generally you are not building WebSphere cells everyday


Most work in WebSphere around day to day administration


Installing applications


Defining resources


Helping application developers to solve their problems etc etc


2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

26

Separation of product and configuration data


On distributed typically


Configuration data that defines a cell


Stored under config sub
-
directory


Which is located under root directory where product code
installed



On z/OS


Product code stored in one file


Configuration data stored in different file



Advantages


Easy to manage multiple versions of WebSphere


And even multiple versions at multiple maintenance levels


Easy to change a cell to run on a new maintenance level


And to fall back to previous maintenance level

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

27

What it looks like…

LPAR A

ZFS

/maint
-
level
-
n

/maint
-
level
-
n+1

/dmgr
-
config

/node
-
1
-
config

/was
-
product

ZFS

ZFS

ZFS

ZFS

LPAR B

/maint
-
level
-
n

/maint
-
level
-
n+1

/node
-
2
-
config

/was
-
product

WAS product code

Mounted Read Only


ZFS


Can only be
mounted to one
LPAR for WRITE


Directories


Can be set up as
shared across
LPARs in a sysplex

See techdoc: http://www
-
03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/WP100396

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

28

IBM TechDocs Web Site


gold mine of info


IBM Techdocs


Many articles related to WebSphere on z/OS


http://www
-
03.ibm.com/support/techdocs/atsmastr.nsf/Web/WhitePapers


http://www
-
03.ibm.com/support/techdocs/atsmastr.nsf/Web/Techdocs


Search using keyword of WebSphere




Full course and lab notes on how to setup WebSphere on z/OS
available


V6.1


http://www
-
03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/PRS2202


V7


http://www
-
03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/PRS3422


2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

29

And now for the big difference…


A WebSphere server on Windows and Unix


Is one JVM



On z/OS


A WebSphere server split into two components


Control Region


A JVM


Handles receiving requests and sending the response


One or more Servant Regions


A JVM


Where the application code runs

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

30

z/OS


Server structure and inbuilt
clustering

Java Virtual Machine

9080

HTTP

9443

SSL

Threads

T

T

T

T

A Started Task

Java Virtual Machine

Threads

T

T

A Started Task


Servant Region

WLM

http://www.abc.com/context/app

One logical WebSphere Server

Java Virtual Machine

Threads

T

T

A Started Task


Servant Region

Control Region

R

R

R

R

WLM detects requests waiting and goal not being met!

Starts a second servant

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

31

Why is server split asunder?


Control region


Runs authorised code


Has access to restricted z/OS functionality


Handles HTTP/S communications



Servant region


Does not run authorised code


Just application code


Means application code cannot get access to
authorised z/OS services


Prevents application code being used to attack
the system

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

32

The Control Region is watching


Control region


For each request


Records time dispatched


If no reply within specified timeout period


Kills the servant region


WAS V7


Introduces more advanced options



Why does request not complete within timeout period?


Typically some backend system not responding


Could also be that application is looping



On distributed


what would happen in such a case?


Nothing until someone intervenes



On z/OS


Servant cancels results in automatic restart of new one


If second servant already running, then it take new requests
immediately


2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

33

Adjunct Server


WebSphere servers can be configured with a System
Integration Bus ( SIB )



On distributed


This runs inside the server



On z/OS


This runs in a separate servant







Because running as a separate process


Can be less need to create separate WAS servers just to run SIB’s

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

34

Clustering on distributed


Feature of WebSphere


Is clustering



On Windows and Unix


Horizontal clustering
implemented by having number
of servers


Often on physically
separate servers



Each server independent of the
other


No sharing of resources


No inherent common
management


Typically remote from
corporate data



Windows/Unix Server 1

DMGR

Windows/Unix Server 2

Server

Node Agent

Windows/Unix Server 3

Server

Node Agent

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

35

Clustering on z/OS


Logically the same


Still have nodes, clustering etc



But to implement with z/OS need a Parallel Sysplex



Parallel Sysplex


Provides system wide management of multiple z/OS
environments


Each z/OS system runs in a Logical PARtition ( LPAR )


AIX now has this term, about 15 years after created on System z


Resources such as CPU, I/O shared between LPAR’s


Looks like this…


2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

36

Parallel Sysplex

Each z10


has 64 CPUs


each 900 MIPs approximately

Two z10


act as Coupling Facilities

Approx CPU MIPs = 30 * 64 * 900 = 1728000 MIPs

Total Memory = 30 * 1.5Tb = 45Tb

Parallel Sysplex


run on 1 to 32 z10’s

DASD

Farm

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

37

High Availability

LPAR A

AppServer

CR

SR

Node Agent

CR

LPAR B

AppServer

CR

Node Agent

CR

Coupling Facility

MQ

DB2

Shared Queues

MQ

DB2

Cluster

Queue

Sharing

Group

DB2 Data

Sharing


WebSphere and other IBM products designed to take advantage of
Parallel Sysplex


DB2 Data Sharing


All data available across the sysplex, no partitioning


MQ Shared Queues


Stored in Coupling Facility


Messages available to all QMGRs, and thus to applications in WebSphere

SR

SD

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

38

Heterogeneous Cells

IBM Techdoc: http://www
-
03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/WP100644

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

39

WAS V7 on z/OS only features


Following features added to WAS V7


To take advantage of z/OS capability

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

40

In the past if a timeout value expired because of a “hung” thread, the servant
region was abended. New function attempts to “shake loose” thread.

Request

1.
Request received

2.
Timer value set

3.
Thread dispatched

4.
Time pops

a.
Pre
-
V7
--

Abend Servant with EC3

b.
V7.0
--

Attempt “Thread Hang Recovery”

If the thread
can

be freed, then it is and the user request is terminated

(with a choice of what “dump action” is taken
--

none, svcdump, javacore, heapdump, and traceback)

If the thread
can not

be freed, then other processing occurs:


Servant may still abend


(if settings give WebSphere no other options)


Or the thread may be left in hung state if threshold value not exceeded


(a new optional variable sets a percentage of hung threads that causes abend of servant)


Thread may be left in hung state if last “surviving servant”


(a new optional variable prevents last servant from abending when minimum servants more than one)



Result: helps to avoid restarts of servant processes

Improving the existing WebSphere runtime

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

41


WebSphere high availability


Relies on the components of a cell “chatting” to each other over
TCPIP


On distributed, not noticed since distributed servers typically not
sharing CPUs


On z/OS, was noticed



WebSphere for z/OS V7


now allows XCF to be used as the protocol for high availability
discovery and failure


Implemented by configuration change


Via Admin GUI or wsadmin


Can still use TCPIP as distributed does


Allows more realistic ‘interval’ for failure determination



Aims to reduce CPU usage compared to heartbeat mechanism



Leveraging existing z/OS technologies

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

42

Application Server

Dynamic Cache

External
Cache Group

FRCA

Exploitation of FRCA is really as an extension to the existing Dynamic Cache
(“DynaCache”) capability of each Application Server.

FRCA is defined as an “External Cache Group,” and the “Adapter Bean” is
what provides the function to access FRCA.

Application servers > [server] > Dynamic cache service > External cache group

Enable AFPA (one of four options
for external cache member)

Provide a port on which AFPA will
listen (this is a net
-
new port usage)

Check the “Enable FRCA” box

Provide FRCA settings

Leveraging existing z/OS technologies


Fast Response Cache Accelerator ( FRCA )


a function of TCP


exploited by HTTP Server for many years


a very good caching mechanism


WAS V7 servers may now exploit it as well

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

43

Improving the existing WebSphere runtime


z/OS uses mechanism called System Management Facilities ( SMF )


To record detailed information about processes running on z/OS



WebSphere V7 for z/OS provides new SMF record 120 Subtype 9


Provides more information about WebSphere transactions


Helps customers with chargeback information, such as:


What ran (Application, Servlet, EJB method, MDBean )


When it ran, how long it took


Who ran it (Calling host:Port, Security ID: Origin, Received, Invocation)


Resources used (CPU


CPs, zAAPs, zIIPs, Bytes transferred)


Other (Classification Names)


Overhead significantly less compared with current SMF Type 120
(subtypes 1
-
8) records.


A browser to display the contents of the new SMF records is provided


2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

44

WAS z/OS Optimized Local Adapters


Statement of direction



IBM currently intends to deliver an Optimized Local Adapters feature for
WebSphere Application Server for z/OS V7 as a future enhancement.
This feature would provide non
-
WAS applications (written in Cobol,
C/C++, HLASM) running on z/OS with a native connection optimized for
exploitation of WebSphere Application Server for z/OS. The connection
would be specifically optimized for application aware collocation within a
single z/OS image where performance requirements justify closely
bound integration between the application service calls. The target
availability is second quarter 2009.



The information on the new product is intended to outline our general
product direction and it should not be relied on in making a purchasing
decision. The information on the new product is for informational
purposes only and may not be incorporated into any contract. The
information on the new product is not a commitment, promise, or legal
obligation to deliver any material, code or functionality. The
development, release, and timing of any features or functionality
described for our products remains at our sole discretion
.


2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

45

WAS z/OS Optimized Local Adapters


Infrastructure and interfaces that provide local connectivity into and out of
WebSphere


Infrastructure


New optimized channel within WAS


Supports inbound work load management (WLM)


Secure connection that propagates context


Interfaces


C/C++, PL/I, Cobol, and HLASM interfaces for calling WAS


Java classes based on JCA for calling into CICS


For use were WebSphere in
the same LPAR as target



Performance


Inbound: 5x better than Web
services


Support for large messages
(>32K)


= WOLA V1.0

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

46

WAS on z/OS


The good life


Aspects of running WebSphere on z/OS that
make life easier

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

47

Can mix 32 and 64 bit servers in the cell

WebSphere Server

Node Agent

32 bit JVM

Deployment Manager

Node

Node

WebSphere Server

Node Agent 64 bit JVM

Node

Cluster

WebSphere Server

9085

HTTP

9445

SSL

32 bit JVM

WebSphere Server

9086

HTTP

9446

SSL

32 bit JVM

Cluster

9080

HTTP

9440

SSL

32 bit

JVM

9085

HTTP

9445

SSL

64 bit

JVM

9085

HTTP

9445

SSL

64 bit

JVM

LPAR A

LPAR B

Cell

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

48

Sysplex Distributor

http://10.1.2.9:9080/context/app

LPAR A


10.1.2.3

Control

9440

SSL

JVM

Servant

JVM

TCPIP

9080
-

LPAR A, B

9440
-

LPAR A, B

Sysplex Distributor

10.1.2.9

LPAR B


10.1.2.5

Control

9440

SSL

JVM

Servant

JVM

TCPIP

9080

HTTP

9080

HTTP


Part of z/OS


Load balances
requests


Based on WLM


Works at TCPIP
level


Cannot maintain
affinity


Fast failover


Less then 1
second

9080
-

LPAR A, B

9440
-

LPAR A, B

Sysplex Distributor

10.1.2.9

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

49

Close/Open TCPIP Ports


Can issue command to tell server to stop listening


And then to start listening again


From SDSF


F <server>, PAUSELISTENERS


Result:


No longer accepts TCPIP and IIOP requests, or reads MQ
msgs


F <server>,RESUMELISTENERS


Result:


Server starts accepting TCPIP and IIOP requests again, and
reading MQ Messages


No Server restart


Only available on z/OS



Use:


Deploy new version of application without having to stop/start
server

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

50

Cryptographic support


zSeries hardware


Supplied with cryptographic hardware cards



WebSphere on z/OS


Able to take advantage of these cards


http://publib.boulder.ibm.com/infocenter/wasinfo/v7r0/index.jsp?topic=/com.ib
m.websphere.zseries.doc/info/zseries/ae/csec_sslkeystoreconfs.html



2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

51

DMGR


run on any LPAR

LPAR A

DMGR

CR

SR

AppServer

CR

SR

Node Agent

CR

LPAR B

DMGR

CR

SR

AppServer

CR

SR

Node Agent

CR

ZFS


Whitepaper
-

WP101140



How to make DMGR mobile
across LPARs


http://www
-
03.ibm.com/support/techd
ocs/atsmastr.nsf/WebInde
x/WP101140

ZFS

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

52

Modify command

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

53

WAS V7


Display thread details


Display command can provide thread information


http://publib.boulder.ibm.com/infocenter/wasinfo/v7r0/index.jsp?topic=/com.ibm.websp
here.zseries.doc/info/zseries/ae/rxml_dispthreads.html


2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

54

Trace via modify command


On z/OS there is another way to set trace
dynamically



Can issue modify command to the server


F WZSR01A,tracejava='com.ibm.*=all‘


Can set trace back to defaults


F WZSR01A,traceinit



2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

55

WebSphere V6.1 startup times

STC Name

Z9
-

1 CPU

Z9
-

2 CPUs
+ 2 zAAPS

Z10
-

1 CPU

Z10
-

2
CPUs + 2
zAAPS

WZDEMN

0:00:01

0:00:01

0:00:01

0:00:01

WZDMGR

0:01:48

0:01:13

0:00:52

0:00:36

WZDMGRS

0:01:21

0:01:02

0:00:27

0:00:17

WZNODE5

0:00:56

0:00:30

0:00:27

0:00:16

WZSR01A

0:01:43

0:00:59

0:00:53

0:00:33

WZSR01AS

0:01:25

0:00:48

0:00:29

0:00:19

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

56

Production on one box


A large Australian financial company


Runs WebSphere V6.1 on z/OS


On two LPARs


On one z990 machine


24x7


Peak rate over 200 requests/sec


In production since 2004



Would you feel comfortable running the same


On a single Windows server?


On a single Unix server?

2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

57

Summary


We’ve compared


WebSphere on z/OS with


WebSphere on distributed



Seen what’s the same


And what’s different



Discussed unique features available in WAS on
z/OS



Hopefully you agree that like thunder and lightening


WAS on z/OS


Is not so frightening


2009 WebSphere Services Technical Conference

© 2009 IBM Corporation

58

Questions


Please remind students to complete
their session evaluation form



Thank you!!