Given
this, routing of a set of queries consists of two steps: first, the scheduler decides in
which order to route the queries in the input queue. Second, it routes the individual
queries in this order.
Classification of Query Routing Strategies
We can classify query routing strategies alongside the following two dimensions:
??? Query-dependence:.As first classification, we can distinguish whether routing
strategies make the routing decision query-dependent or -independent. Many
conventional routing approaches are query-independent, which means that the
query to be routed does not influence the routing decision. In contrast, more
sophisticated approaches base the routing decision on the current query to be
routed.
??? Previous.knowledge:.Routing strategies can be further distinguished with
regard to flexibility whether the routing decision is done dynamically at runtime
or based on precomputed data. Standard affinity-based routing is a good
example for a routing algorithm using precomputed affinity data. In contrast,
cache approximation routing uses a completely dynamic approach.
In the following, we give an overview of existing query routing strategies and relate
them to the presented classification scheme.
Conventional.Routing.Strategies
Traditionally, routing algorithms are published under the term transaction routing
that can be considered equivalent to query routing as typically only read transactions
are considered.
Pages:
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452