45 YEARS

RS
SERVING THE GLOBAL
FORTUNE
UNE 500

RULES PROCESSING OPTIMIZATION FOR

MAINFRAME DATACENTERS
Or, How In-Memory Technology Can Help You
Retain and Increase Market Share

A DATAKINETICS WHITE PAPER

Rules Processing Optimization for Mainframe Datacenters

Table of Contents
Transaction processing on the mainframe 3
Challenges for rule processing 3
The business solution 4
Business rules embedded in mainframe code 4
Mainframe-based OTS Rules Management 5
Distributed systems-based OTS Rules Management 5
Mainframe-based in-memory rules management 5
Mainframe-based in-memory rules management: A closer look 6
De-coupling business rules 6
Modifying in-memory business rules using modern technology 6
Conclusions 7

DataKinetics Data Performance & Optimization | 50 Hines Road, Suite 240 Ottawa, ON, Canada K2K 2M5 | © 2023 DataKinetics 2
Rules Processing Optimization for Mainframe Datacenters

Transaction processing on the mainframe

Mainframe systems are the core business processing engines for virtually all of the world’s
biggest banks and credit card companies, and most of the world’s biggest insurers,
retailers, and healthcare organizations. Despite the myriad of recent disruption technologies
like analytics, cloud, Big Data, mobile and IoT, the mainframe remains the core.

Challenges for rule processing

Rules management is on the critical path – it defines how a business interacts with its
customers, and for this reason, business rules are among the organization’s most valuable
intellectual properties. For speed of processing, business rules are often embedded within
mainframe COBOL applications – see Figure 1.

The reason for this is that all of the rules processing is handled by the
transaction processing applications; they do not need to access them Legacy Rules
from a repository, like a database. Changing business rules in this Processing Application
environment typically involves not only people with business expertise,
but also mainframe application developers, managers, DBAs, etc. The
mechanism comprises application recompiles, testing, staging, and
deployment on the productions systems – and for this reason, in many
cases, rule changes can take weeks or even months.
The business need is to take business rules management out of the
mainframe development environment because business rules MUST be Embedded Business Rules
easy to maintain. More specifically, an organization MUST be able to
change / create rules quickly – preferably in matter of hours or even Figure 1: Embedded business rules
minutes – to facilitate the business being agile, and able to react very
quickly to changing business conditions.

It’s not necessary to take either rules processing or rules management off the mainframe, but
it would be prudent to take business rules management out of the hands of the mainframe
development groups. Their processes are generally not conducive to rapid code changes
– and while that may not be an issue for most aspects of mainframe development, it is not
effective for business rules creation and updates.

If your competition can react faster to changing market conditions than you can, your
business will lose market share, and will have great difficulty winning new business. You need
applications that are market-agile. Churning out business rule updates on a development
schedule just won’t cut it.

DataKinetics Data Performance & Optimization | 50 Hines Road, Suite 240 Ottawa, ON, Canada K2K 2M5 | © 2023 DataKinetics 3
Rules Processing Optimization for Mainframe Datacenters

The business solution

A modern rules management approach will give you the tools needed by your business
experts to react quickly to their market analysis. An optimized solution will include:

• Efficient business rules management

• A business rules repository
• Fast business rules processing
Your priorities will determine the best possible solution for you; if rules processing (and
transaction processing) speed is secondary, you might consider moving off the mainframe.
If that’s too much, too soon, you could consider a phased approach that will allow an
ordered migration at a later time. Alternatively, if you want to stay on the mainframe, you
could keep transaction processing on the mainframe and migrate the rules processing
portion to a cheaper platform, using a standard COTS (Commercial, Off The Shelf) rules
processing solution.

On the other hand, if rules processing speed is important, you would almost certainly need
improved rules management AND the absolute best processing performance. If this is the
case, you should consider a high-performance, low-maintenance, market-agile mainframe
rules processing solution.

Let’s dig a little deeper at some of the solutions open to you:

• Business rules embedded in mainframe COBOL (or C, Java, etc.) code

• Typical mainframe-based COTS rules management
• Typical distributed systems-based rules management
• Mainframe-based in-memory rules processing

Business rules embedded in mainframe code

The embedding of business rules within application code was born out of necessity decades Users initiate
ago – and it is still today the fastest possible technique for running business rules in real MIS change request
time. The benefits are many, but the main ones are that there is no I/O required to access
business rules, and there is no database overhead code path. This technique has been, is, MIS/user meetings
and always will be the fastest possible way to run business rules logic.
MIS priority schedule
But the technique comes with some very serious drawbacks – new business rules
development – or even the simplest changes to existing rules logic – require code changes.
Change program
Figure 2 shows the rules change management process in some detail.

As shown, users – in this case finance Product Managers, Business Managers, Marketing Recompile linkedit
Managers, etc. – make requests based on the marketing needs of the organization, or
based on changing market conditions. This initiates a meeting between MIS (management Test
information services, a.k.a., IT) and the users. This meeting might actually be a regularly
scheduled MIS meeting – one for which the users have to wait, because the required MIS User acceptance
personnel aren’t otherwise available.

At this point, the MIS would prioritize the work required to make changes to the applications. Production
The bad news is that it may not automatically be their top priority; a particularly hard pill
to swallow for folks who really believe that they’re dealing with business-critical concerns.
Figure 2: The process
Once the work can begin, the established development process takes over – code updates, to change code-based
program recompiles, multiple layers of testing, debugging, acceptance (typically involving a business rules
staging effort), and finally, going into production systems.

This whole process is what has to change.

DataKinetics Data Performance & Optimization | 50 Hines Road, Suite 240 Ottawa, ON, Canada K2K 2M5 | © 2023 DataKinetics 4
Rules Processing Optimization for Mainframe Datacenters

Mainframe-based OTS Rules Management

Off-the-shelf rules management systems provide precisely what is missing in the embedded-
code solution – a completely seamless rules management capability. Business rules run
on the mainframe (leveraging the fast processing, highly secure, high-throughput capacity,
high reliability), and allow for business rules changes in days, or even hours.

These types of solutions – like IBM’s ODM (Operational Decision Management) or CA’s Aion
Business Rules Expert – are the best for many users. However, there are some very serious
drawbacks – rules processing is MUCH slower than embedded rules processing speed. So
that lightning-fast rules processing that you had with embedded business rules will now
take much longer to process – by an order of magnitude.

Further, this type of rules processing consumes much I/O, and takes a long code path to
business rules – if they’re running on the mainframe, you can expect a significant increase
in the mainframe operations cost – that monthly bill.

However, for some businesses, these drawbacks may not matter.

Distributed systems-based OTS Rules Management

Distributed off-the-shelf rules management systems are far more popular than similar mainframe-
based solutions. Once an organization has decided that the speed of processing business rules
is not a business-critical need, they will quickly learn that distributed-based solutions running
on UNIX, Linux or Windows, are far less costly than their mainframe-based counterparts.

These types of solutions – like Oracle Business Rules or SAS Business Rules Manager –
command the lion’s share of market share in this area. Like their mainframe counterparts,
these solutions are considerably slower than any mainframe-based solution, often consume
considerable network traffic and take long code paths to business rules, in scenarios where
mainframe transaction processing applications are driving them.

But again, there are businesses that may not be adversely affected by longer business rules
processing times, so this solution may be perfect for them.

Mainframe-based in-memory rules management

This type of solution is ideal for organizations relying on the mainframe for uber-fast
transaction processing, who have a requirement for both fast business rules processing
AND seamless & efficient business rules management. The benefits are: an extremely short
code path to business rules, and virtually no I/O consumed for rules processing. All that is
needed is an efficient API to allow your applications to get at rules information.

Optimization of legacy applications can be implemented piecemeal— you can start Users initiate internal
by selecting only those areas causing the most pain, and leaving the rest of the legacy change request
application unchanged. And the end-game is in-memory applications that are practically as
fast as embedded rules applications. The main difference is that they are easy to maintain. Edit tables
Figure 2 shows us the reason for the weeks/months timeline for in-program-based rules
management. In-memory rules processing can help improve that turn-around time to days/ Test
hours/minutes for rules changes.
Production
Figure 3 shows how this process completely bypasses the development organization – in fact,
the entire process can be handled by the non-technical side of the business. Users can makes
changes directly to the rules database without the need for technical assistance. In fact, the last
Figure 3: The process to
two parts of this process may not even be needed, depending upon the implementation details.
change memory-based
Mainframe high-performance in-memory technology is the best solution for large business rules
organizations requiring both fast rules processing (it processes rules almost as fast as the
embedded rules technique), and efficient and fast rules management.

DataKinetics Data Performance & Optimization | 50 Hines Road, Suite 240 Ottawa, ON, Canada K2K 2M5 | © 2023 DataKinetics 5
Rules Processing Optimization for Mainframe Datacenters

Mainframe-based in-memory rules management: A closer look

The key to in-memory rules processing is the sharply reduced code path – see Figure 4.
Calling Application

SSAS RDS DM BM VSAM

Med Mgr

DASD

BSDS Logs SQL Parse SQL Opt Rec Map Index Mgr DSM Buffer Pools LRW IRLM

tableBASE
tableBASE Driver

Figure 4: The shortest path to your code

The top part of Figure 4 shows the typical DBMS code path, even taking in-memory
buffering into account. The bottom part of Figure 4 shows the ultra-short high-performance
in-memory code path. In general, your business rules will be stored in Read-Only in-memory
tables. For that reason, in-memory technology can help your applications bypass some of
your database overhead when accessing table-based business rules.

De-coupling business rules

In-memory technology allows business rules to be de-
coupled and externalized from rules processing and
transaction processing applications. As shown in Figure
5, transaction processing is handled by rules processing
applications – they typically get data from DASD (not
shown here), but using the short code path, they can
access some data from in-memory tables. The added Business rules externalized Transaction
benefit is that business rules can be modified without into High-performance Processing
touching the rules processing applications. in-memory tables
Rules Processing
Business managers would use in-memory table handling Applications
applications to modify rules, create new rules without
involving software development teams at all. In this way, Figure 5: De-coupled business rules
rules changes that once required weeks / months to
implement now require only days / hours / minutes.

Modifying in-memory business rules using modern technology

In-memory technology has been around for decades – and the tools used to modify business
rules were accessed using a green-screen dumb terminal. But it’s not 1989 anymore, and
green screen operations are definitely not the preferred technique. So there are better
solutions for business managers.

DataKinetics Data Performance & Optimization | 50 Hines Road, Suite 240 Ottawa, ON, Canada K2K 2M5 | © 2023 DataKinetics 6
Rules Processing Optimization for Mainframe Datacenters

A straight-forward upgrade on the rules management side makes the in-memory rules
processing solution even more powerful. Using modern tools with very little configuration
setup and a small amount of setup programming, business rules can be managed in real
time. Rules authoring and management can be accomplished by non-technical personnel
(Business Managers, etc.), who can make changes to business rules literally in minutes.
Windows-based Rules
Authoring/Management Application

Web Interface

Rules Database Custom Integration In-memory Business Transaction

Application Rules Processing

Rules Processing
Applications

Figure 6: Modern rules authoring and management

The custom integration application can be as simple as a translation from MS Excel or .CSV
to the in-memory table format. This can be accomplished by one of several mainframe/web
services integration services solutions that are in use today.

Conclusions
Mainframe processing for the biggest banks, insurers and credit card processors is deeply
entrenched, and is not going to change for years to come. The mainframe continues to
provide the best possible computing performance for these organizations. Those mainframe
systems are running their business rules processing at lightning speed - but managing
business rules is becoming an acute challenge.

In today’s fast-paced and fluid business environments, businesses need to react as quickly
as possible to changing market requirements - and outdated rules management processes
can no longer get in the way of that.

The best solution for these businesses is mainframe-based in-memory rules management.
This technology can help them to take the technical lead over their competitors – and to
take market share from them using in-memory market-agile applications.

It allows for seamless and quick rules management AND extremely fast rules processing -
the best of both worlds.

© DataKinetics Ltd., 2023. All rights reserved. No part of this publication may be reproduced without the express written permission of DataKinetics Ltd.
DataKinetics and tableBASE are registered trademarks of DataKinetics Ltd. Db2 and z/OS are registered trademarks of IBM Corporation. All other trademarks,
registered trademarks, product names, and company names and/or logos cited herein, if any, are the property of their respective holders.

DataKinetics Data Performance & Optimization | 50 Hines Road, Suite 240 Ottawa, ON, Canada K2K 2M5 | © 2023 DataKinetics 7