Complete
We recommend Client/Server Software Architectures as prerequisite reading for this technology description.
Since 1994 mainframes have been combined with distributed architectures to
provide massive storage and to improve system security,
flexibility,
scalability, and
reusability in the
client/server design. In
a mainframe server software architecture, mainframes are integrated as servers
and data warehouses
in a client/server environment. Additionally, mainframes
still excel at simple transaction-oriented data processing to automate
repetitive business tasks such as accounts receivable, accounts payable,
general ledger, credit account management, and payroll. Siwolp and Edelstein
provide details on mainframe server software architectures see
[Siwolp 95,
Edelstein 94].
While client/server systems are suited
for rapid application deployment and distributed processing, mainframes are
efficient at online transactional processing, mass storage, centralized
software distribution, and data warehousing
[Data
96]. Data warehousing is information (usually in summary form) extracted
from an operational database by
data mining (drilling down into the information through a series of related queries). The purpose of data warehousing and data mining is to provide executive decision makers with data analysis information (such as trends and correlated results) to make and improve business decisions.
Figure 20: Mainframe in a Three Tier Client/Server
Architecture
Figure 20 shows a mainframe in a three tier client/server architecture. The combination of mainframe horsepower as a server in a client/server distributed architecture results in a very effective and efficient system. Mainframe vendors are now providing standard communications and programming interfaces that make it easy to integrate mainframes as servers in a client/server architecture. Using mainframes as servers in a client/server distributed architecture provides a more modular system design, and provides the benefits of the client/server technology.
Using mainframes as servers in a client/server architecture also enables the
distribution of workload between major data centers and provides disaster
protection and recovery by backing up large volumes of data at disparate
locations. The current model favors "thin" clients (contains primarily user
interface services) with very powerful servers that do most of the extensive
application and data processing, such as in a
two tier architecture. In a three tier client/server
architecture, process management (business rule execution) could be off-loaded to another server.
Mainframes are preferred for big batch jobs and storing massive amounts of
vital data. They are mainly used in the banking industry, public utility
systems, and for information services. Mainframes also have tools for
monitoring performance of the entire system, including networks and
applications not available today on UNIX servers
[Siwolp 95].
New mainframes are providing parallel systems (unlike older bipolar machines)
and use complementary metal-oxide semiconductor (CMOS) microprocessors, rather
than emitter-coupler logic (ECL) processors. Because CMOS processors are
packed more densely than ECL microprocessors, mainframes can be built much
smaller and are not so power-hungry. They can also be cooled with air instead
of water
[Siwolp 95].
While it appeared in the early 1990s that mainframes were being replaced by
client/server architectures, they are making a comeback. Some mainframe
vendors have seen as much as a 66% jump in mainframe shipments in 1995 due to
the new mainframe server software architecture
[Siwolp 95].
Given the cost of a mainframe compared to other servers, UNIX workstations and personal computers (PCs), it is not likely that mainframes would replace all other servers in a distributed two or three tier client/server architecture.
Mainframe technology has been well known for decades. The new improved models
have been fielded since 1994. The new mainframe server software architecture
provides the distributed client/server design with massive storage and
improved security capability. New technologies of data warehousing and data
mining data allow extraction of information from the operational mainframe
server's massive storage to provide businesses with timely data to improve
overall business effectiveness. For example, stores such as Wal-Mart found
that by placing certain products in close proximity within the store, both
products sold at higher rates than when not collocated.1
By themselves, mainframes are not appropriate mechanisms to support graphical
user interfaces. Nor can they easily accommodate increases in the number of
user applications or rapidly changing user needs
[Edelstein 94].
Using a client/server architecture without a mainframe server is a possible
alternative. When requirements for high volume (greater than 50 gigabit),
batch type processing, security, and mass storage are minimal, three tier or two tier architectures without a mainframe server may be viable alternatives. Other possible alternatives to using mainframes in a client/server distributed environment are using parallel processing software architecture or using a database machine.
A complementary technology to mainframe server software architectures is
open systems
. This is because movement in the industry towards interoperable heterogeneous software programs and operating systems will continue to increase reuse of mainframe technology and provide potentially new applications for mainframe capabilities.
This technology is classified under the following categories. Select a
category for a list of related topics.
|
Name of technology
|
Mainframe Server Software Architectures
|
|
Application category
|
Client/Server (AP.2.1.2.1)
|
|
Quality measures category
|
Maintainability (QM.3.1)
Scalability (QM.4.3)
Reusability (QM.4.4)
|
|
Computing reviews category
|
Distributed Systems (C.2.4)
|
Darleen Sadoski, GTE
Frank Rogers, GTE
10 Jan 97 (original)
1
Source: Stodder, David. Open Session Very Large Data Base (VLDB) Summit,
New Orleans, LA 23-26 April, 1995.
