Lazy-Merge: A Novel Implementation for Indexed Parallel K-Way In-Place Merging

In this section, we discuss why Lazy-Merge, a nonstandard merging format, is advantageous, in terms of a better run-time, over the standard format. We support this advantage using experimental results obtained from applying our standard to a real-time application, and later, we present its applicability to a general case. We considered merging e-mail lists to show that LazyMerge has an advantage over the standard merging format. In the merging of e-mail lists, there is a set of e-mail lists, and we assume that each e-mail appears in only one e-mail list. Each e-mail list contains a set of fields for each record, i.e., first name, last name, and e-mail address. Merging e-mail lists is a common task in which copies of e-mail lists are merged into a single list with respect to a certain field; we assumed the e-mail address field is the one to consider for the merging process. The original e-mail lists are kept untouched for further use, whereas copies of the lists are involved in the merging task. Thus, the final merged list is ordered by e-mail address. k e-mail lists merging can be realized as k-way merging, where each e-mail list is a segment. In some cases, there is a need to analyze the merged list contents to know the share of each list in a certain portion of the list. For example, if sending a letter succeeded only for the first 10% of the final merged e-mail list, then the user might need to know how many e-mail addresses from each e-mail list received this letter. This can be viewed as the share of each segment for a certain portion of the final merged list.