D
dbeavon
Guest
>> about 3 times higher on the networked appserver hmmm... You say 3 times higher, and Progress claims there will be a 300% improvement after fixing "client/server" connectivity in the OE 12 database servers (now with support for threading and joins). That seems very coincidental. Perhaps they are using your numbers to predict the potential performance improvements that their customers will experience. I'd guess that any ABL code which uses thousands of "find firsts" will NOT see a 300% overall improvement, even if the bottleneck is entirely due to the database. And moreover it seems hard to believe that they can predict that kind of improvement in real-world applications. I haven't seen OpenEdge v.12 in action myself, but I'll be very relieved if it gives us a 50% improvement, let alone 300%. I may be a bit cynical but it seems the only way you can guarantee that someone will see a substantial performance improvement is if you were removing a "pause" statement from the original code. ;-) @jankeir can you please tell me what you've done (in 11.3) to improve the performance for nested loops? (eg. for each order, each order line)? Do you try to break those apart into separate & independent FOR-EACH-NO-LOCK operations? Otherwise in the case of the nested FOR-EACH it seems that the client/server performance degradation is more dramatic and is proportional to the number of orders.
Continue reading...
Continue reading...