References in periodicals archive ?
For TPC-D, the knee in its miss ratio curve occurs at less than 1%, but its miss ratio continues to improve at buffer pool sizes beyond 10% of its data size.
On the other hand, TPC-D's LRU miss ratio tends to be much closer to its MIN miss ratio than the production workloads.
Since the TPC-D queries are run serially in the power test, this shortcoming should not affect TPC-D.
TPC-C's total page reuse at almost 60 is the highest among all the workloads, while TPC-D's reuse at about 17 ranks among the lowest.
The only exceptions are for TPC-D and to a lesser extent, Bank.
For TPC-D, a 16-entry run buffer is more appropriate; this size will be used in the rest of the article.
Results presented above show that the various workloads, especially TPC-D, contain some very long sequential runs.
Figure 16 summarizes the average prefetch hazard for the production workloads and for TPC-D. The individual plots for the production workloads can be found in Hsu et al.
Observe further from Figure 17 that TPC-D's miss ratio is reduced by 50%, which is well within the spectrum defined by the production workloads.
Figure 19 presents the average prepurge hazard for the production workloads and for TPC-D. The individual plots for the production workloads can be found in Hsu et al.
Acronyms browser ?
Full browser ?