Spatial Disk Defragmentation, Google, and 64 Bit Cuil
Spatial Data on a Disk
Either a virus or a bad VPN driver filled up my disk with garbage files. In the process of cleaning things up my disk became very badly fragmented.
From the geodatabase’s perspective though, the file may still be fragmented. When I zoom in close on a large featureclass, the features that fall within the extent might be scattered in many different places on the disk, even though the file containing them is not fragmented.
How much would performance improve if features that are geographically near each other were placed on the same disk clusters?
Spatial Data in Memory
Still, it would be better to avoid disk access altogether. I have MSDN library loaded on my hard drive, however I find it faster to use Google to search msdn online. I suppose Google is faster since most info I’m looking for is already loaded into memory somewhere. Getting data from memory in Googles’ computers thousands of miles away is still faster than getting data from my local disk. As far as I know the Google Platform uses 32 bit computers – constraining them to 4 GB memory. Perhaps Google is faced with an Innovators Dilemma in deciding when to make the jump from 32 bit to 64 bit? If Cuil can educate Wall Street on the 64 bit advantage maybe they can compete with Google.
Stephen Arnold traces much of Cuil’s advances — and the advances of other search engines — to simple hardware designs that were carried out years ago by Alta Vista, the pioneering search engine build around the Alpha 64-bit processor at Digital Equipment Corporation. – Information Week