Re: SMP and 7.x performance questions?

This is a multi-part message in MIME format. --------------36C27634C9 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cheryl Kendricks wrote: > > Everyone keeps repeating Round Robin is not a Good Fragmentation Method. > Well in my Class I was told different. So, Before I start this process > could someone give me the Good, Bad and Ugly of using Round Robin. You will find in life in general that one solution is rarely the best for all situations. This is as true with fragmentation schemes as with anything else. The major disadvantage to RR is that fragment elimination cannot be done. As there is no "rule" to apply or consider regarding what rows are placed in a particular fragment, when looking for data sequentially you will have to search all the fragments. On the other hand, if you are accessing the data via an index that is not fragmented RR (i.e. is not an attached index on a RR fraged table), then the problem goes away. If the index IS attached to a RR fraged table, then you will have to search each of the index fragments when doing an index lookup. If you have a lot of cpus available, this may not be the worst thing in the world as the lookups can be done in parallel. However if you are using expression-based fragmentation, and your queries are structured such that fragmentation elimination can be done, you will generally see better performance, as there will be less work to do to satisfy the query. The upside to RR is that it produces a perfectly even distribution of data across the fragments. The biggest problem with expression-based fragmentation is finding an expression that gives you an even distribution of rows across the fragments. Quite often, an expression that would provide such is costly and often unuseable for fragment elimination (a mod() of a column, for example). So users are frequently faced with the choice of using RR to get even distributions and having to do scans of all fragments, or using expression-based fragmentation and getting fragment elimation but suffering with uneven distribution of data across the frags. To make a blanket statement that either method is universally good or bad simply displays a lack of knowledge or experience - each will have its advantages and disadvantages in different situations. A better-yet solution will be offered in the 8.X release of Online, in the form of a hash-based fragmentation. This will provide the distribution attributes of RR with the potential for fragment elimination as in expression-based. -- Dave Kosenko, Informix Professional Services **************************************************************************** While it is true that there is more than one way to skin a cat, the cat himself generally fails to appreciate the differences. --------------36C27634C9 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="IFXDISCL.TXT" ************************************************************************* Note: please do not send me email asking about features, or asking about Informix problems and how to solve them. I answer what questions I can in this forum (comp.databases.informix) when I have the time to spare. For questions on features, call your local sales rep or check out the Informix web site (http://www.informix.com). For technical problems, call Informix tech support. ************************************************************************* Disclaimer: All opinions expressed in this message are well-reasoned and insightful; needless to say, they are not those of Informix Software, its partners or lackeys. Anyone who says otherwise is itching for a fight. --------------36C27634C9--