NonStop Systems Introduction

Requirements of ZLE Systems
NonStop Systems Introduction—527825-001
2-4
Consolidated View of All Data
performance. Two frequently used measures of performance are response time and
system throughput.
A widely accepted definition of response time is the elapsed time between user input
to a system and the user’s receipt of confirmation that the transaction is complete.
This measurement usually includes the time it takes the data to travel across a
communications line, the time it takes the computer to access the database, and the
time it takes the computer to send a response. At the completion of the transaction,
the result of that transaction would be available throughout the entire enterprise. Most
applications in a zero latency environment will require sub-second response time.
An example of response in an ATM application is the number of seconds that elapses
between the time the user enters a personal ID and specifies the amount of cash to be
withdrawn and the time that the machine delivers the money.
An area of particular concern these days is that of business over the internet.
Increasingly, both business-to-business (B2B) and business-to-consumer (B2C)
transactions are taking place over a company’s Web site. In the internet environment,
customers expect immediate service; delays are simply not tolerated. In fact, studies
have shown that customers will leave a Web site if they must wait longer than two
seconds for a response. Therefore, it is vitally important that a ZLE system provide for
fast, efficient internet operations.
System throughput refers to the number of transactions that a computer system can
process in a given period. It is often defined in terms of transactions per second
(TPS). A ZLE system typically must support workloads on the order of thousands or
even tens of thousands of transactions per second.
Furthermore, in today’s business environment it is not enough for a system to provide
a high level of performance. The system must provide that performance at a
reasonable cost. The price/performance (the ratio of price to performance) rating
determines the number of simultaneous users who can be supported by a specific
hardware configuration for a specific cost. Hence, price/performance is a critical
measure of a ZLE system.
Consolidated View of All Data
Typically, a company’s data is stored in local databases and passed across a
spiderweb of networks. Passing data from one system to another places a heavy
burden on the systems and results in high latency. Moreover, the data in the
operational databases might be redundant or inconsistent.
The network of systems shown in Figure 1-1 on page 1-5 is extremely inefficient. The
data store described in Section 1, Introduction, is designed to solve this problem. The
data store is a central repository of data that provides a persistent (that is, always
available) cache of information that business systems can leverage, instead of passing
data across a spiderweb network.
By integrating the data into a central storage cache, it can be synchronized across
multiple enterprise applications and can be readily accessed by all applications