http://ayende.comAyende @ RahienCopyright (C) Ayende Rahien 2004 - 2021 (c) 202660Martin, It is right next to show query results, in the bottom of the screen. http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment8http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment8Mon, 28 Jun 2010 07:44:36 GMTWhere can I find the query plans window in NHProf? Can't find it. http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment7http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment7Sat, 26 Jun 2010 09:52:24 GMTSimon, Theoretically, yes, you could. The problem is that I am updating Lucene in the background, and that are some questions that I have to have a transactional answer to The last update one, for example, is important for replication http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment6http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment6Sun, 20 Jun 2010 21:07:34 GMTAyende, "How do you find all the documents that were changed after a certain point in time" Just thinking out loud, but if your Storage had that knowledge (ie, LastUpdated) and Lucene was hooked deeply to it, then Lucene would know about it, wouldn't it? Just to make it clear, I find it VERY interesting that you are implementing your own indexing engine, I'm just wondering what could be done with Lucene in this particular case. http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment5http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment5Sun, 20 Jun 2010 20:50:32 GMTSimon, Indexing is a bit more complex than a binary choice between Lucene or not. For example, when implementing the managed storage for Raven, I need to store documents in the disk, and then I need to read them back. How do you find document "users/ayende" ? That is what I refer to as Index Seek. How do you find all the documents that were changed after a certain point in time? That is an Index Scan. Those aren't tasks that I can shell to Lucene. http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment4http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment4Sun, 20 Jun 2010 20:38:30 GMTDid you evaluate keeping Lucene as the index engine with your new Transactional Storage? If so, maybe you could write a post about that, explaining why (it seems) you chose to write your own indexing engine on top of your storage instead of just optimising the storage. http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment3http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment3Sun, 20 Jun 2010 20:28:55 GMTZvolkov, This is pretty much because I am focusing on just internal details on what this actually matter. I am using index scans quite often, actually. Good point http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment2http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment2Sun, 20 Jun 2010 15:13:44 GMTYou seem thinking only about finding one entity, but I'll tell you what else the index scans are useful for: to be able to enumerate entities in a pre-sorted way. Either to extract a pre-sorted sublist of the entire list, or to retrieve the entire list presorted. This could be useful as is (because for example the query has an ORDER BY) or because another part of the query could benefit from presorting (e.g. a merge join). http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment1http://ayende.com/4544/table-scans-index-scans-and-index-seeks-on-my#comment1Sun, 20 Jun 2010 15:07:38 GMT