This nine minute demo by Dave Williams of CICS Development in Hursley, UK demonstrates some of the performance improvements made in CICS TS V4.1 when using the new Sysplex Optimized WLM function. Throughputs for similar workloads are compared between CICS TS V3.2 and CICS TS V4.1 with Sysplex Optimized WLM switched on and off.

Just to give a little extra background, Dave was the chief developer and does an excellent job here of clearly describing the new improvements.

A slight confession, I wrote the tool the video uses to analyse and chart the workloads. It is an internal IBM test tool, so you can’t get your hands on it, but if you would like to know more or you think it would be useful to you, let me know.

If you are interested in seeing more videos like this, direct from the development team, add a comment, with maybe a sentence or two about what you would like to hear about.

Notes

• For more details on Sysplex Optimized WLM see here.
• Dave used the term ‘Base table’ in the clip. A ‘Base Table’ is more commonly known as a Resource Table. The new table Dave referred to is WLMATARG.
• The tests performed here show an artificial workload running 10,000 transactions using the Distributed Routing mechanism. The results you achieve with your workloads in your environments may differ.

What I’m going to try and do in this post is to (hopefully) explain some of the reasons why the WUI behaves as it does.

History

Originally CICSPlex SM was never going to have a WUI at all. Circa CICSPlex SM 1.3 a GUI was being looked at. As this was the mid to late 1990′s the platform for this GUI was OS/2! Yes O.K. I’ll admit it, I am an OS/2 fanboy err old-bloke – I still use it every day (please don’t tell my manager), but this decision was nothing to do with me. Due to various issues, the OS/2 based GUI was never shipped and instead a Web Based Interface aka WUI was developed.

By the time the WUI was ready, it was shipped as a no-charge feature on CICSPlex SM 1.4 – An exclusive element of CICS TS 1.3.

The prereqs for the web browser were kept very low, e.g. HTML 3.2 with frames, tables and cookies. Talk to some web designers (or Chris) and they will probably give you a stern look for mentioning the frame or table words.

Questions

Why does the WUI open a new window on signon?

The WUI relies on every request from the web browser being sent back to the server. The WUI checks for this by using a sequence number in it’s interactions with the browser. If the sequence number is not what the WUI expects, you will receive a ‘Navigation error’ response in the WUI.

The easiest way to get one of these errors is to use a browser forward or back button. To discourage users from using these buttons, the signon process causes a new window to open, and on this request the WUI asks the browser not to include the browser navigation controls. Although this decision might be considered heresy today, and it does not stop users from using keyboard shortcuts, mouse operations, or putting the browser controls back, it was done to try and discourage users from using these controls.

So the reason the ‘Begin signon’ button opens a new window, is simply to get a window without browser navigation buttons. As this response to the original CICS-L question indicates, you can bypass the ‘Begin signon’ by tweaking the URL that you use. Note however although this works, it is not an intended ‘interface’, so it potentially could change in the future (not that I know of any plans to do so).

Why the strange tab behavior?

The WUI knows absolutely nothing about tabs. They weren’t around when the WUI was developed originally and no explicit support for tabs has ever been added to the WUI.

The only thing the WUI does when it opens a new window for Help or the InfoCenter is to use the target=’_blank’ HTML option on the anchor tag. How the browser interprets this in conjunction with the ‘Open in new Tab’ options you use to get it to go to a tab is completely up to the web browser.

Why do I get another ‘Begin Signon’ window?

So you already have a ‘Begin signon’ tab, why do you get another one if you refresh a WUI window after you have signed off? Well it is simply because you no longer have any valid credentials on the WUI server, so the WUI starts the signon dialog. The WUI has no idea of what windows/tabs your browser has open (once you have signed off).

Requirements are important

So none of this ‘fixes’ the weirdness, but I wasn’t trying to provide a fix, just explain why some of these things happen. Could the experience be better? Sure. If there are things that really bug you, get in those requirements, either via organisations such as SHARE, or talk to your IBM account team, who should be able to submit requirements on your behalf. Although it might be frustrating waiting for those requirements, if they aren’t on the potential ‘To-Do’ list within the Lab, that dramatically lowers the chances of getting things changed.

CICSPlex SM WUI Data interface

This connection is available with all currently supported CICS TS releases. It requires that the customer has setup CICSPlex SM and a CICSPlex SM WUI (Web User Interface). CICS Explorer users configure a connection to the usual WUI hostname and port (that you would use in a web browser).

This interface is ‘read only’. You can retrieve data, but not ‘alter’ it, for example, while you can see CICS tasks, you cannot purge them.

As this is a CICSPlex SM interface, you can work with the context and scopes that are normally available, exploiting the CICSPlex SM ‘Single System Image’ (SSI) capabilities.

CICS Management Client Interface (CMCI)

This is the connection that has been keeping my fellow blogger, Chris, busy. It is new with CICS TS V4. It provides full ‘CRUD’ (Create / Remove / Update / Delete) capabilities, so unlike the Data Interface option previously described, you can purge tasks, discard files etc.

The CMCI can operate with two different types of host:

CICSPlex SM WUI

This option requires CICSPlex SM. Using a CICS TS V4 WUI, there is a new WUI System parameter (CMCIPORT) to allow you to configure an additional port number that is used for the CMCI interface. A WUI can now have two TCPIPSERVICEs and ports, one for HTML based WUI web browser users, and another for the XML based CMCI users – which includes CICS Explorer.

As this is a CICSPlex SM interface, you can work with the context and scopes that are normally available with the CICSPlex SM ‘Single System Image’ (SSI) capabilities.

System Managed Single Server (SMSS)

SMSS allows the CMCI with a standalone CICS region. You don’t need to configure any CICSPlex SM components, or set up any CICSPlex SM infrastructure like CMASes and WUI Servers.

Using this variation of the interface, you can only work with the single CICS region that you are connected to. Each CICS, that you manage, needs to run the SMSS code, and you have to have a TCPIPSERVICE and unique port number configured for each region.

Examples

The following graphic attempts to show some different CMCI topologies. With CICS Explorer, the SM perspective is connected via one connection at a time. So to work with any of the green CICS in the CICSplex, the Explorer user connects to the CMCIPORT on the WUI server on the left and CICSPlex SM handles the requests to all of the CICS that are part of the CICSplex, e.g. CICSPlex SM communications will be used to manage the green CICS on the right.

The brown CICS in the graphic above, attempt to show that the CICS Explorer user must connect to each of these CICS in turn to manage the ‘single system’.

With the CMCI, a CICS can either be CICSPlex SM managed, or an SMSS. You can’t use both interfaces at the same time. Only one of the ‘pink arrows’ can be active at a time.

Getting to CICS TS V4

If you are a CICSPlex SM user today with CICS TS V3, you can get up and running to look (and not touch) with the SM perspective very quickly – using your existing WUI.

Depending on your configuration, as soon as you get a CICS TS V4 CMAS (MP) and WUI server running, you may be able to start performing actions on your CICS TS V3 CICS while you complete your migration to CICS TS V4. However There are some rules you need to follow:

1. The CICSPlex SM Maintenance point CMAS must be the highest level, so CICS TS V4.
2. The CICSPlex SM WUI Server must be the highest level and connect to a CICS TS V4 CMAS.
3. The CICSPlex SM code in a CMAS and WUI Server must run on a CICS 660 region.
4. The CICSPlex SM agent code running in a user CICS (MAS) needs to be the same as CMAS release.

The CICSPlex SM V4 code can run in older releases of CICS, so you can manage a CICS 640 region with CICSPlex SM V4, and use the CMCI to gain CRUD access (for supported resources) to this ‘CICS TS V3 CICS’.

CMASes can be different releases, so for example, you can have a CICSPlex SM V4 CMAS communicating to a CICSPlex SM V3 CMAS, but see rule 1 above, the V4 CMAS must be the MP CMAS for all the CICSplexes that are to be accessed. Using this setup, you will have CRUD access to the CICS TS V3 CICS managed by the CICSPlex SM V3 CMAS.

The following graphic attempts to illustrate:

Depending on the other CICS tools installed, you may have support for additional types of connection, but this post is just attempting to describe the capabilities of the SM perspective.

If you have CICS regions to manage or even just monitor, CICSPlex SM provides an ideal environment for getting your job done.  With CICS TS 3.2 it now has a much improved installation that is integrated directly with CICS, so it has never been easier to try it out.  Better yet, you don’t have to convince anyone to “show you the money” as it is bundled right there alongside CICS.

So why use CICSPlex SM?  What makes it so great?  Well here are 6 things, that I think makes CICSPlex SM a must have!

1. Single System Image (SSI)
   CEMT is a great transaction for keeping an eye on your regions but it can become a bit of a chore to run CEMT on each, even if you only have 3 regions running.  Using SSI, CICSPlex SM pulls all of your regions together into a CICSplex that you can manage and monitor as if they were a single system.  For example, you can run CEMT-equivalent queries CICSplex-wide in one request.  Using CICSPlex SM’s powerful criteria and filter mechanism you can take it a step further and make requests like discarding all programs, CICSplex-wide with a name matching “PRG*” (starting with “PRG”).  SSI is not just limited to CEMT like operations, infact it provides huge amounts of power when combined with the other reasons.
2. Business Application Services (BAS)
   Slap a pair of underpants on the outside of CEDA’s clothes and sew a great big S on it’s chest and you have BAS.  Simply, BAS is a resource definition and deployment manager but as with all things CICSPlex SM, it does it on a much bigger scale.  Instead of using the CICS CSD, BAS uses it’s own data repository, that we affectionately call the DREP (catchy eh?).  From BAS you can define any CICS definition you like and you don’t need to put them into groups to install them (yes, installation CICSplex-wide in one request).  That’s pretty cool in itself but BAS is all about applications.  Individual resources are all well and good but BAS lets you define and deploy entire applications faster than a speeding bullet (ok this is just a rough guess, but a quick straw pole suggests a top speed of somewhere between that of a speeding bullet and the CICSPlex SM team heading for a team breakfast – thanks Neil! ).  Define your resources, put them into logical groups of your choosing (multiple applications can share resource groups) and define your application to CICSPlex SM.  BAS will install both your remote and local definitions for you into the scopes you specify (again CICSplex-wide).  This literally scratches the surface of what BAS can do.
3. Workload Management (WLM)
   So now you have deployed your application, you want to make sure it absolutely flies and it’s availability is assured.  CICSPlex SM WLM is just the tool that you need.  At it’s simplest, just tell WLM where your routing regions and target regions are (remember to connect them) and WLM will start efficiently routing your work in an optimal manner.  If you have an affinity to a particular system (say you have GETMAIN’d some shared storage on a region and while a particular user is signed on you need to keep routing them to that region) don’t sweat, with WLM you can also separate a particular part of your workload to handle this sort of affinity and others.  Using the WLM queue algorithm WLM can balance your systems by analysing the health, link, load and affinities, so you can be sure we will choose the right regions to maximise your throughput.  Using the WLM goal algorithm, WLM will do everything queue does but it can also take your z/OS WLM goals into account as well when making routing decisions.  If availability is your goal the WLM is king, even if the region managing your workload goes down, your workload can continue to function as normal until is recovered.
4. Real-Time Analysis (RTA)
   Now everything is up and running, you probably want to make sure everything stays that way.  Sometimes knowing about a problem, 5 minutes after it happens is about 10 minutes too late.  With RTA you can quickly set up monitoring points of interest and provide instant real-time notifications to either NetView, the CICSPlex SM API or even out via our Web User Interface (WUI).  Not only does RTA provide you with notifications but it can also correct the problem for you and naturally it does this CICSplex-wide.  RTA can perform modifications to resources when an event is notified, for example re-enabling a file when it was disabled by that trainee system programmer who keeps clicking on things he shouldn’t (you know the one).  RTA can provide System Availability Monitoring (SAM) for keeping an eye on the health of your systems (going SOS, dumping, load is getting high), MAS Resource Monitoring (MRM) for keeping and eye on your resources and Analysis Point Monitoring (APM), which is pretty much the same as MRM but limits the number of events you will receive if you are monitoring cloned or similar systems.
5. Application Programming Interface (API)
   So all that power is great but it is nothing without some rock solid ways of accessing it.  The first of these is through the CICSPlex SM API.  If you have any experience writing EXEC CICS commands the EXEC CPSM commands that CICSPlex SM provides should feel at least syntactically familiar to you.  With the API, you can manage resources (create, delete, discard, install, perform actions, set attributes, filter and copy) and even listen for RTA events.  Again, all of this is CICSplex-wide.  You can run the API as part of a normal CICS application or even in batch or TSO.
6. Web User Interface (WUI)
   The second of the interfaces mentioned here, and pretty much the main entry point to CICSPlex SM for most users, is the Web User Interface (WUI).  I should point out that there are also another couple of really nice interfaces (BATCHREP and EYU9XDBT) but you’ll have to rely on the infocenter for more information on those right now.  The WUI provides a single point of control for all of this CICSPlex SM power.  It sits on top of the CICSPlex SM API and the ever awesome CICS WEB API to provide a clean, fast and super-quick browser interface into your CICSPlex SM world.  A WUI is not limited to just a single CICSplex either.  It will happily show you information about any CICSplex you have specifically given the WUI access to.  Everything you can do in CICSPlex SM is available through the WUI and fully RACF-enabled too!

So that is a whirlwind tour of why you should consider giving CICSPlex SM a shot. I normally run somewhere between 2-10 regions at any time and I can’t remember the last time I saw a CEDA or CEMT screen.

So if you decide to give CICSPlex SM a chance, let us know how you get on, we are always looking to improve on the CICSPlex SM experience and we can only do so with your honest feedback.  Tell us what you need to start using CICSPlex SM today. (Sadly we don’t have any free z10′s to give away but I’m sure we have some more stickers under my desk that could do with a new home for the best answers).

CICS has a number of different facilities that let you share files. For example function shipping, shared data tables and RLS. However a slight snag in the design of the WUI object service meant that it was not possible to use any of these facilities and maintain integrity of the EYUWREP file.

In a nutshell, the WUI object service may use multiple physical records in the EYUWREP file to represent one logical record, for example a view set may contain any number of views. It is the view set that is the ‘container’ for the views, so is saved and loaded as a single entity.

The WUI does have a locking service, but this were built on the assumption that each WUI server had its own unique EYUWREP file. So the locking service only provides for locks within a single WUI server. This is one of the reasons why it is documented that the EYUWREP file can NOT be shared between WUI servers.

Prior to CICSPlex SM V3R2M0, there were no checks on the EYUWREP file definition to ensure that it had not been changed to RLS or a remote file or a data table. We didn’t expect customers to change the IBM supplied definition, but some of them did anyway.

When these customers (with ‘modified’ EYUWREP file definitions) migrate to CICSPlex SM V3R2M0, then the WUI server will fail to initialise with message:

EYUVS0916E FILE EYUWREP has invalid value: YES for option: RLSACCESS

This message is issued in V3R2M0 because we made some changes to the way CICSPlex SM resource definitions are built. Bootstrap definitions are created during CICS startup, and then if the bootstraps are invoked, the remaining definitions are created on the fly. Because we acknowledged that some people change definitions, we allow user definitions, but we now check the definitions to make sure it is something vaguely close to the IBM supplied one. If not, the EYUVS0916E message is issued.

So even though you might have a ‘modified’ EYUWREP definition and the modification was for something that was documented as not being supported due to the risk of corrupting data, a change has recently been made to CICSPlex SM V3R2M0 by APAR PK61740.

The PTF for this APAR changes the WUI locking service so that it can provide protection for the logical EYUWREP records. This allows the EYUWREP to be shared within a sysplex between multiple WUI servers via RLS. Note that that we are only supporting RLS for the sharing, not function shipping, or shared data tables etc, just RLS.

So take a look at this APAR if you want to share WUI server data repositories. Please make sure you check out the hold info in the PTF cover letter for further details