Re: Long-running queries

In <3v5pl7$15i@excepcion.cc.gatech.edu> gukal@cc.gatech.edu (Sreenivas Gukal) writes: >Thanks a bunch to all the people who have responded (either personally >or on the net). Most of the replies fall into two categories. >1. Run the queries at less than degree 3 isolation. >2. Maintain a copy of the database and run queries on it. >The first one seems fine, if the result need not be transaction >consistent. Looks like quite a few users can live with that (or >forced to live with that :-). I see a couple of problems with >the second solution. Its okay to maintain a copy if your database >is small. But once you start having terabyte or larger databases, >the disk overhead and the copying time can become considerable. >Besides the queries do not see the current data. Not many companies run terabyte databases. There must be a way to break down a data set that is so large. It isn't like you would be doing a terabyte copy every few hours - the transactions just update the mirror normally. >Now for the thesis issues :-) A couple of recent research papers >try to solve this by using transient versioning. The basic >idea here is that every update creates a new version. The queries >access the prior versions while the transactions update the >current versions. Storage overhead is reduced by making the prior >versions transient; they are removed when the queries accessing >them complete. My work goes on similar lines, but also shows >that the storage overhead for transient versions can be further >reduced. "KEEP IT SIMPLE STUPID" Sorry I had to say it. This sounds way too complex. It is much easier to spend the money on a good mirror database. There is a lot of overhead needed to maintain that transient versioning too. If you want to do it right you gotta mirror or run stuff at night. >I confess, all my knowledge is from books and journals. I did >not really know if this problem is important in real world. >From your responses, this looks like a real problem. Please feel >free to respond if you have any thoughts on this. Thanks once >again for your time. >-- >-------------------------------------------------------------------------------- >Sreenivas Gukal Ph: 404-853-9390 > email: gukal@cc.gatech.edu >-------------------------------------------------------------------------------- -- Hal Cooper - hal@teleport.com| The opinions and statements contained in this MIS Manager | posting are the sole responsibility of the Paper Products Marketing | author and have not in any way been reviewed _____________________________| or approved by my employer.