Amdahl versus Google

Amdahl worked at IBM and then founded his own companies, including Amdahl and Trilogy. Interestingly, he was arguing against expecting too much from parallel processing, which at the time—the mid-1960s—was beginning to attract substantial interest. He argued for “the continued validity of the single processor approach and of the weaknesses of the multiple processor approach in terms of application to real problems,” basing his analysis on the proportion of computation that was typically sequential at that time, such as data management housekeeping.4

His argument may have been true at the time, but, today, loosely coupled parallel processing is used to great benefit in a variety of important computing tasks. One that we all use several times each day—without necessarily realizing or appreciating—is online search.

Suppose that you were in a large public library, say, with 1 million books, without any Web access or online catalog. Moreover, suppose that you needed to find the exact book and exact page number with a specific unusual phrase. You might get lucky and, accidentally, immediately pull the correct book off the shelf and turn to the right page. Generally, it would take quite a while to read through each book in search of the text fragment. In fact, you might be extremely unlucky and find that whatever order you picked meant that you were extremely unlucky and searched through all 1 million books before you found the phrase. On average, for a given random ...

Get Cloudonomics: The Business Value of Cloud Computing, + Website now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.