The importance of mainframe performance

It’s so easy to forget, or just take it as read, that mainframes have been able to successfully run with five nines availability for well over a decade. What that means is achieving 99.999% of scheduled uptime. In other words, it means that unscheduled downtime is less than five and half minutes in a year! Now that kind of amazing performance is something that boxes running other operating systems can only dream of. Some are working towards that level of availability, but others (you know who I’m thinking of here) aren’t even close.

But I wasn’t thinking about performance in that sense. We just take it for granted the operating system is going to be always working. What I was thinking about was the performance of the major subsystems running under z/OS. It’s very important to take steps to ensure that CICS or IMS are performing optimally. Monitoring software can be installed that will identify when preset thresholds are reached. They help identify bottlenecks and then the appropriate action can be taken to resolve them. This is the kind of stuff systems programmers have been working away at for years. They’ve been using faster processors, faster I/O, more efficiently-coded transactions, until every component is working as well as it can.

In previous blogs, we’ve talked about monitoring software that can arrange for alerts to be sent to designated staff as text messages or e-mails – allowing them to access the nearest iPad or laptop and take steps to resolve the new problem.

As well as CICS and IMS, there are monitors for DB2, WebSphere, and z/OS itself. These can all be integrated and produce wonderful moving graphs or other displays that allow users to tell at a glance whether everything is OK or whether a slight tweak to the subsystem is required. In addition, we’ve had software that learns how to maintain high performance and make appropriate changes on-the-fly, without any human intervention.

But the problem that many sites now face is what they can do if, for example, IMS users are reporting slow response times, but the problem appears to be coming from outside the IMS subsystem rather than from inside it. For example, what appears to be an IMS performance problem could be a CICS, DB2, WebSphere, or z/OS performance problem. The challenge facing systems programmers in this situation is to correlate performance data in IMS with activities in these other systems in order to discover the cause of the slow response time.

One new solution is Transaction Analysis Workbench, which is an IMS tool. If you’re interested in how to approach this type of situation, how to gather the necessary information from multiple subsystems, and then analyse, diagnose, and resolve the problem, you’ll be interested in the webinar from the Virtual IMS user group this week.

The Virtual IMS user group runs free webinars every other month. During the webinar, a technical expert shares their hard-won knowledge with the rest of the group. The webinars use Citrix GoToMeeting, which means you don’t have to face the hard task of convincing your company to fund your user group experience – you just sit down at your laptop and log in.

Anyone wishing to join the webinar needs to join the user group – which is also free. The next meeting is at 10:30 Central Standard Time on Tuesday 8 February. The user group’s Web site (where you can join) is at www.fundi.com/virtualims.

Cloud and the future of mainframes

CA Technologies released a survey on Wednesday entitled “Mainframe as a Mainstay”. The survey was conducted on 200 senior level US-based mainframe executives by Decipher Research. Amongst the results was the information that 73% of respondents confirmed that the mainframe is – or will be – part of their organization’s cloud computing strategy. The question posed at CA’s recent webinar was whether that result came as a surprise.

Now, as you may know, I’m a big fan of cloud computing. In fact, I’ve an article about cloud computing and IMS in the current issue of zJournal. It struck me that these results were very much in line with the recent results, also published by CA, from a survey carried out by Vanson Bourne, a market research company based in the UK. They conducted more than 300 interviews during August with European IT decision makers. Their report was called “Mainframe - The Ultimate Cloud Platform?”. They found a slightly higher figure of 79% of organisations believing the mainframe is an essential component of their cloud computing strategy. The also found 70% of respondents agreeing that cloud computing will sustain or extend the mainframe environment.

On the other hand, only 10 per cent of mainframe sites in BMC’s survey in October said that using their System z machines to run cloud computing or SaaS applications was an important priority for them in the coming year. Quite a difference! Similarly, my own straw poll at the Guide Share Europe conference at the beginning of November, in an IMS session, found that no-one seemed interested in cloud computing. I think that reflects real-life economics in that they were very much focused on what was available now that would make the business run better and their lives easier – how they could do more with less.

I think what we’re seeing is a difference between the attitude of mainframe staff, who want to get the job done with the tools available now and the pressure of fewer staff etc, and senior managers who are looking more strategically towards the next step.

The Arcati Mainframe Yearbook 2011 user survey has a couple of questions about cloud computing. It will be interesting to see the results from that. And, by-the-way, if you haven’t completed a survey yet, you can do so by going to www.arcati.com/usersurvey11.

Among the survey’s other findings we see:

• A majority (80%) responded they will be maintaining or increasing spend on mainframe staff in the next 12-18 months.
• More than three-quarters (76%) will maintain or increase their investment in mainframe software during the next 12-18 months.
• Nearly half of respondents (46%) are looking for industry leadership from vendors on the evolving role of the mainframe in the enterprise.
• 61% of respondents don’t believe the IT industry does enough to promote mainframe career opportunities to recent graduates.
• 35% believe that recent graduates are not as proficient in mainframe technology as their counterparts that entered the workforce 10-years ago.
• 61% said that hiring either took much longer than expected, took long enough to negatively impact their IT operation or are still looking for talent after more than six months.

Picking up on the 35% of respondents who believe recent graduates are not as technology proficient as their counterparts that entered the workforce 10-years ago, webinar panellists were asked whether this was something they were seeing in the workforce.

I think the truth is that even the ancient Greeks felt that youngsters weren’t as good as they used to be! Certainly when I started working on mainframes, we were a mixed bag of youngsters, and many of those less capable or that-way-inclined left – leaving the enthusiasts and the highly technically-capable. I assume the criticism can always be applied. New people at any job just aren’t very good. And once they are quite good, they’re promoted to a different one.

We hear the mainframe referred to as a dinosaur – even though we know signs point in the opposite direction – in addition, the mainframe has a reputation as older technology with a middle-aged workforce, so the panellists were asked whether the fact that 52% of those surveyed cited Facebook and LinkedIn as the most effective recruiting tools came as a surprise.

My first response whenever a sentence includes the words mainframe and dinosaur is to point out that dinosaurs ruled the Earth for 160 million years. Humans have existed for say 200,000 years. Draw your own conclusions!

I’m not at all surprised that middle-aged people are using Facebook and LinkedIn and many other examples of social media. At the end of 2008, the answer may have been surprising, at the end of 2009 it may have surprised some people, but not at the end of 2010. These are IT people we’re talking about – of course they’re going to know what’s going on in the cyber world. I also imagine, next year, that figure will be much higher than 52%. Mainframers know about social networking. Look how many of them blog and are on Twitter.

You can find more information about the survey at www.ca.com/us/news/Press-Releases/na/2010/CA-Technologies-Survey-Reveals-Mainframes-Role-as-Anchor-in-Cloud.aspx.

IMS logical relationships

The latest webinar from Virtual IMS CONNECTION was entitled, “The logical answer: IMS Logical Relationships”, and was presented by Aurora Emanuela Dell’Anno, an Engineering Services Architect with CA. Aurora is an IMS and DB2 Specialist Systems Engineer, with a past in application development. Aurora has spent the past 14 years specializing in various aspects of the IBM software world, with particular attention to database management and data warehousing. Keeping pace with the times, the past years have been dedicated to in-depth work on, and study of, data warehousing realities, trouble-shooting systems installations, and application performance issues – this included getting her hands dirty with distributed platforms, customers’ implementations, and problem-solving needs in this field, all from the perspective of systems engineering and database administration. Aurora works for CA in the highly-focused Mainframe Solutions Center.

Aurora started her presentation by explaining exactly what an hierarchical database was and what relational data structures were. She then asked why IMS needed logical relationships and suggested the reason was to access segments by a field other than the one chosen as the key, or to associate segments in two different databases or hierarchies. Aurora informed the user group that IMS provides two very useful tools to resolve these data requirements: secondary indexes logical relationships.

Thinking about database design, it can be useful to normalize the data. This helps to break data into naturally associated groupings that can be stored collectively in segments in a hierarchical database. It also breaks individual data elements into groups based on the processing functions, and groups data based on inherent relationships between data elements.

The database types that support logical relationships are HISAM, HDAM, PHDAM, HIDAM, and PHIDAM. There are no logical relationships with Fast Path DEDB or MSDB.

Logical relationships are formed by creating special segments that use pointers to access data in other segments. The path between the logical child and the segment to which it points is called a logical relationship.

There are three types of logical relationships:

• Unidirectional
• Bidirectional physically paired
• Bidirectional virtually paired.

A unidirectional relationship links two segment types (logical child and its logical parent) in one direction. A one-way path is established using a pointer in the logical child.

A bidirectional physically paired logical relationship links two segment types, a logical child and its logical parent, in two directions. A two-way path is established using pointers in the logical child segments.

A bidirectional virtually paired logical relationship is like a bidirectional physically paired relationship. It:

• Links two segment types, a logical child and its logical parent, in two directions, establishing a two-way path
• Can be established between two segment types in the same or different databases.

A logical child segment exists in only one database. When going from database A to database B, IMS uses the pointer in the logical child segment. When going from database B to database A, IMS uses the pointer in the logical parent, as well as the pointer in the logical child segment.

To establish a logical relationship, three segment types are always defined – the physical parent, the logical parent, and the logical child.

Aurora explained that the pointers used in logical relationships fall into two categories – direct and symbolic. We can implement logical relationships using both types, however, a direct pointer is a true pointer.

There are four types of pointers:

• Logical parent (LP)
• Logical Child (LC)
• Logical Twin (LT)
• Physical Parent (PP).

LP pointers point from the logical child to the logical parent. An LP pointer is in the prefix of the logical child and consists of the 4-byte direct address of the logical parent.

Logical child pointers are used only for logical relationships that use virtual pairing. With virtual pairing, only one logical child exists on DASD, and it contains a pointer to a logical parent. The logical parent points to the logical child segment.

Two types of logical child pointers can be used:

• Logical Child First (LCF)
• A combination of LCF and Logical Child Last (LCL) pointers.

Because LCF and LCL pointers are direct pointers, the segment they are pointing to must be in an HD database. The logical parent (the segment pointing to the logical child) must be in a HISAM or HD database. If the parent is in a HISAM database, the logical child must use a symbolic pointer to point to the parent.

A symbolic LP pointer consists of the Logical Parent’s Concatenated Key (LPCK). It can be used to point into a HISAM or HD database.

Physical Parent (PP) pointers point from a segment to its physical parent. IMS generates PP pointers automatically for HD databases involved in logical relationships.

Logical Child Pointers are used only in logical relationships with virtual pairing. When virtual pairing is used, there is only one logical child on DASD, called the real logical child. This logical child has an LP pointer. The LP pointer can be symbolic or direct.

Two types of logical child pointers can be used:

• Logical Child First (LCF) pointers
• A combination of logical child first (LCF) and Logical Child Last (LCL) pointers.

CA’s Aurora Dell’Anno went on to inform the user group that logical twins are multiple occurrences of logical child segments that point to the same occurrence of a logical parent segment. Two types of logical twin pointers can be used:

• Logical Twin Forward (LTF)
• A combination of LTF and Logical Twin Backward (LTB).

HALDBs (PHDAM, PHIDAM, and PSINDEX databases) use direct and indirect pointers for pointing from one database record to another database record. The use of indirect pointers prevents the problem of misdirected pointers that would otherwise occur when a database is reorganized. The repository for the indirect pointers is the indirect list data set. The misdirected pointers after reorganization are self-healing using indirect pointers.

The relationship between physical parent and logical child in a physical database and the LP pointer in each logical child creates a physical parent to logical parent path. For a physical parent to logical parent path, the logical parent is the destination parent in the concatenated segment.

For a logical parent to physical parent path, the physical parent is the destination parent in the concatenated segment.

When use of a physical parent to logical parent path is defined, the physical parent is the parent of the concatenated segment type. When an application program retrieves an occurrence of the concatenated segment type from a physical parent, the logical child and its logical parent are concatenated and presented to the application program as one segment. When use of a logical parent to physical parent path is defined, the logical parent is the parent of the concatenated segment type. When an application program retrieves an occurrence of the concatenated segment type from a logical parent, an occurrence of the logical child and its physical parent are concatenated and presented to the application program as one segment.

In both cases, the physical parent or logical parent segment included in the concatenated segment is called the destination parent.

There must be one physical DBD for each of the databases in a logical relationship. All statements are coded with the same format used when a logical relationship is not defined, except for the SEGM and LCHILD statements. The SEGM statement includes the new types of pointers. The LCHILD statement is added to define the logical relationship between the two segment types. The pointers for use with HD databases must be explicit in the PTR= parameter.

When a logical relationship is used, you must define the physical databases involved in the relationship to IMS. Also, you often need to define the logical structure of IMS since this is the structure the application program perceives.

With HALDB databases, bidirectional logical relationships must be implemented with physical pairing. When loading a new partitioned database with logical relationships, the logical child segments cannot be loaded as part of the load step. IMS adds logical children by normal update processing after the database has been loaded. HALDBs use an Indirect List Data Set (ILDS) to maintain logical relationship pointers when logically related databases are reorganized.

Aurora recommended a White Paper called Converting Logical Relationships to Support Partitioned Databases written by William N Keene – who some of you will have heard speak at user group meetings and who is a regular attendee. The White Paper provides guidance and examples for converting a set of logically related databases from bidirectional, virtual pairing using direct pointers to bidirectional, physical pairing using symbolic pointers.

The presentation contained a lot more detailed information, including some thoughts on performance. Aurora concluded by stating that logical relationships resolve conflicts in the way application programs need to view segments in the database. With logical relationships, application programs can access segment types in an order other than the one defined by the hierarchy, and have a data structure containing segments from more than one physical database.

Membership of the Virtual IMS Connection is free and open to anyone interested in IMS. Webinars are presented free to members. You can join by going to http://www.virtualims.com/register.aspx and completing the form.