Enterprise Data Management for Transaction and Analytical Processing - Benchmarking Transaction and Analytical Processing Systems - page 20

a

C1 C2

O11 O12 S34 O21 S36 S41

1;5 2;2 2;2 1;3 1;O11;1;5 2;O11;2;2 3;O12;1;3 1;9 2;2 3;5 1;O21;1;6 2;O21;2;2 1;O21;1;3 2;O21;3;5

P2 P8 P2 P2 P8 P2 P3 P7 P8 P3 P7 P3 P8

C1 C2

b

O11 O12 O21

1;5 2;2 1;3 1;9 2;2 3;5

P2 S34 P8 S34 P2 S34 P3 S36 S41 P7 S36 P8 S41

1;5 2;2 3;3

1;6 1;3 2;2 2;5

P2 P8 P2 P3 P3 P7 P8

Fig. 2.4 Example hierarchy instances for the customer order hierarchy. ( a ) No order/shipment
mapping. ( b ) Direct mapping of shipments to order lines

Figure 2.4 exemplifies the aforementioned alternatives for the hierarchical
customer order hierarchy. Nodes in the hierarchical model are instances of entities,
e.g., a customer named C1 . Listing 2.1 presents the orders and shipments depicted

in Fig. 2.4 . Two customers C1 and C2 place orders for the products P2;P3;P7;

and . The numbers provided in the order lines and shipment lines represent the

ordered and shipped quantities.
The different modeling alternatives introduce differing levels of redundancy.
While in the alternative in Fig. 2.4 a only the product instances are repeated and
an overview of complete orders and shipments per customer is simply achieved
by scanning the respective subtrees, the alternative in Fig. 2.4 b stores redundant
shipment data for each order line besides the redundant product data, but it simplifies
queries associating order lines with their shipments. For example, it is easy to
determine that the order for product

P8

P3 of customer C2 is split across two

shipments, that is, , but to analyze the contents of an entire shipment,

a scan of the complete subtree of this customer is needed.
Logical relationships were added in IMS/360 2.0 to efficiently handle many-
to-many relationships [ 138 ]. They interrelate segments from different physical
hierarchies building a logical hierarchy. Logically connected physical hierarchies
may constitute a network data structure, although application data is still stored in
one or more physical hierarchies [ 132 ]. Concerning the previous example, logical
relationships can help to avoid the redundant storage of product records: An own
hierarchy for products can be introduced and logical paths from the order line
records of the customer order hierarchy to these product records can be created.

S36 and S41