Once again I take a jump to your sql-server-performance KB, it's been helpfull several times. Here's the challenge for this time : We have this ERP software, which has now been migrated to SQL2000 (sp2). Because of the migration of the db to SQL2000 and the new software-release of this ERP software, I perform a closer followup regarding db-behavior and performance. One of the things that really cought my attention is that there are almost no clustered indexes defined in this database. There are 1400 tables and only 18 of them have a clustering index. From these 18 tables, only 4 are in use. Aparently we only use 428 tables and it's a 6Gb database. All tables do have a primary key defined and most of them have also other indexes defined. Most of the primary keys have been defined on a single increasing number column. I've asked the softwareproducer/vendor as to why they use practicaly no clusterd indexes. their reply : "We've checked it 2 years ago (with sql7 sp2) and did not find sutable solutions for all users, so we didn't implement it." They promote their software is suitable for SQLserver as well as for Oracle. As a general rule of thumb - also mentioned at your webside - I always put a clustered index on a table. One of the reasons is because that's the only way of getting the datapages reorganized by using dbreindex/indexdefrag. I've noticed at some of our own tables in other databases, that (I guess) by way of use, a heap consumes by far more datapages then a clustered table. Because of this extreem example I get a little shaky when I see no clustered indexes have been defined and no documentation is available to found it. Can you provide more technical info about the differerences in the way sqlserver(2000) treads heaps and clusterd tables datapage wise ? In most discussion groups , there is always a finger pointed to the downside of the use of clusterd indexes, but I've seen no technical issue regarding the use heaps. Are there technical downsides with using heaps ? Is the fact of "by default" missing clustered indexes a standard procedure in ERP databases ? Thanks in advance and keep the good work going on for the sqlserver-community. btw : I've already had a response from Brad stating : "There's been plenty of articles in SQL Server magazine and on Microsoft's website, and in many books (like Inside SQL Server 2000) that point that heaps are never good. That's why it is a good reason to add a clustered index to every index (except perhaps, very, very small tables, but even in those cases, I still do). I would have to write an article to cover why this is, although I do describe some of these reasons on my website under clustered and non-clustered indexes. So, if you are looking for a very technical , in-depth discussion of this, the best is found in the book Inside SQL Server. Assuming you are having performance problems with your ERP database, I would take a Profiler Trace of it and run it through Profiler. It will recommend some of the easiest clustered indexes to add, and then from there, I would begin identifying poorly performing queries, finding the tables they hit, and then add appropriate clustered indexes to them. If you haven't done so, join the website's new forum and post your question there for other feedback from members. Brad "