Large volume updates

Help! This question may be in the FAQ. (If there is a FAQ. It's not at rtfm.mit.edu and it's not on comp.answers or news.answers. If it exists, it has expired on my service provider, and I apologize profusely!) We are using Informix 5.04 for a large archival system. Up to 1.5 million data records are staged in an Informix table prior to being archived by our system.[1] Everything was wonderful, until the customer decided we needed to synchronize two separate archives. In essence, given tables a and b, each with a field named key, we need to partition the data (logically or physically). Best of all, they have two types of data co-mingled in each input stream. I'm currently trying to do this using something like: update a set flag = 0; update a set flag = 1 where key in (select key from b where class = 'SPECIAL'); update a set flag = 2 where key in (select key from b where class <> 'SPECIAL'); update b set flag = 0; update b set flag = 1 where key in (select key from a) and class = 'SPECIAL'; update b set flag = 2 where key in (select key from a) and class <> 'SPECIAL'; Yes, at 1.5 million records each in a and b, this is grotesque. I'm fried, jet lagged, exhausted, and feeling like I must be missing something obvious. Does anyone have any suggestions on how to solve this problem with reasonable performance? Just the first update is taking 45 minutes. Any help would be greatly appreciated! -- Jonathan Crawford Ark Technologies, Inc. [1] For the perpetually curious, if you're wondering why we have our own archival system, it's designed for huge data volumes. To take one example, we have a customer who's input stream consists of over 10 million records per day, with a ten-year retention period and five-minute retrieval requirement. No problem. But, it's not a "relational archive."