A principal problem with integrity constraints' use for monitoring dynamica
lly changing database integrity is evaluation cost. This cost associated wi
th performance of checking mechanisms is the main quantitative measure whic
h must be supervised carefully. Based on literature, evaluating an integrit
y constraint cost includes these main components: (i) data amount accessed;
(ii) data amount transferred across network; and (iii) number of sites inv
olved. In distributed databases, where many networked sites are involved, n
ot only amount of data accessed must be minimized but also amount of data t
ransferred across network and number of sites involved. In [Ibrahim H, Gray
WA, Fiddian NJ. SICSDD: Techniques and implementation. In Proceedings of C
onstraint Databases and Applications, Second International Workshop on Cons
traint Database Systems (CDB'97), Delphi, Greece, January 1997, pp 187-207]
, we introduced an integrity constraint subsystem for a relational distribu
ted database. The subsystem consists of several techniques necessary for ef
ficient constraint checking, particularly in a distributed environment wher
e data distribution is transparent to application domain. Here, we show how
these techniques effectively reduce constraint checking cost in such a dis
tributed environment. This is done by analyzing and comparing generated sim
plified forms to respective initial constraints respecting amount of data t
o be accessed, amount of data transferred across network, and number of sit
es involved during evaluation of constraints or simplified forms. Generally
, our strategy reduces data amount needing to be accessed since only fragme
nts of relations subject to update are evaluated. Data amount transferred a
cross network and number of sites that are involved are minimized by evalua
ting simplified forms at target site, i.e., site where update is performed.
(C) 2001 John Wiley & Sons, Inc.