http://blogs.msdn.com/jonathanh/archive/2004/12/13/why-is-msn-desktop-search-so-slow-at-indexing-my-disk-and-other-tips.aspxA quick follow-up to four points, since " Dogfooding MSN Desktop Search " seems to be getting a lot of attention. First, Joe Wilcox of Jupiter Research is disappointed with the initial indexing speed of desktop search, because it took 2.5 hours to indexen-USCommunityServer 2.1 SP1 (Build: 61025.2)http://blogs.msdn.com/jonathanh/archive/2004/12/13/why-is-msn-desktop-search-so-slow-at-indexing-my-disk-and-other-tips.aspx#284221Tue, 14 Dec 2004 09:06:00 GMT91d46819-8472-40ad-a661-2c78acb4018c:284221Jonathan Hardwickhttp://blogs.msdn.com/jonathanh/archive/2004/12/13/why-is-msn-desktop-search-so-slow-at-indexing-my-disk-and-other-tips.aspx#284367Tue, 14 Dec 2004 06:54:00 GMT91d46819-8472-40ad-a661-2c78acb4018c:284367Kunal KundajeI loved the quick response that I got when I posted the problem I was having with Outlook on the MSN Search blog. Looks like a very enthusiastic and committed team of devs. :)http://blogs.msdn.com/jonathanh/archive/2004/12/13/why-is-msn-desktop-search-so-slow-at-indexing-my-disk-and-other-tips.aspx#285328Tue, 14 Dec 2004 09:16:00 GMT91d46819-8472-40ad-a661-2c78acb4018c:285328Frans Bouma&quot;but I do know one way to speed it up — defragment your hard disk first! &quot; <br>8GB in 2.5 hours is 53MB per MINUTE. I think even in PIOmode4 a harddrive can read all the data fully fragmented. Which means: the disk is definitely not the bottleneck. Or is the process doing very silly I/O? like the I/O decision win32 makes when copying a large file: copy 4KB blocks, no matter how much ram is available. If you have 100MB of ram to spare, the indexer can load files of 100MB at once in memory, then start indexing. Already on the Amiga we knew that if you do I/O and processing in parallel with two (2) buffers (one for reading, the other one is processed, what an invention!) you can achieve very fast performance without having the processes block eachother. <br>http://blogs.msdn.com/jonathanh/archive/2004/12/13/why-is-msn-desktop-search-so-slow-at-indexing-my-disk-and-other-tips.aspx#296716Tue, 14 Dec 2004 16:39:00 GMT91d46819-8472-40ad-a661-2c78acb4018c:296716Rod Trent at myITforum.comhttp://blogs.msdn.com/jonathanh/archive/2004/12/13/why-is-msn-desktop-search-so-slow-at-indexing-my-disk-and-other-tips.aspx#306274Wed, 15 Dec 2004 07:02:00 GMT91d46819-8472-40ad-a661-2c78acb4018c:306274JOEL'S BLOGhttp://blogs.msdn.com/jonathanh/archive/2004/12/13/why-is-msn-desktop-search-so-slow-at-indexing-my-disk-and-other-tips.aspx#316161Wed, 15 Dec 2004 22:48:00 GMT91d46819-8472-40ad-a661-2c78acb4018c:316161Jonathan Hardwick [MSFT]Frans - I'm pretty sure that the indexer is NOT doing very silly I/O. If you go watch the Channel 9 video, you'll see Chris McConnell commenting that he's been working on indexers since 1996, and I can attest to the fact that he's plenty smart. There are clearly several variables at work, because I've seen some people say they indexed half of an 80GB disk in less than an hour, and others say they had to leave the thing running all night and it still wasn't finished. I'll also bet that the performance characteristics of each indexer are closed guarded secrets of the major players, so we might not see a full analysis coming from the inside anytime soon. Nothing to stop anyone on the outside running a performance comparison, though (although better read those click-through EULAs first, just to be sure you can publish it! :-&gt;)