D
dbeavon
Guest
Its not bandwidth that is the problem. It is latency and the number of sequential round-trips that are performed. While saving a business document in our ERP , I've regularly seen *thousands* of client-server round-trips to the OE RDBMS database. This is the nature of the OE database technology. Almost everything happens on a row-by-row basis. As a quick check for latency, I'd suggest "ping"ing the server and reviewing the average round-trip times. In order to get a grip on the number of round-trips that are being made, I typically open task manager in windows and watch the increasing number of packets that are sent/received (image): I think your performance problems will grow proportionally to the (latency*unicasts). One trick that might help in troubleshooting is a trick that Progress tech support told me about. You should *always* start your troubleshooting by running your application code on the same local server as the database - but when you do this, you must make sure to use "-H and -S" parameters for client-server connectivity - instead of using "shared-memory". This trick will take the physical network out of the equation altogether (while leaving the OE client-server technology layer in place). The performance of this configuration must be measured as a "baseline". Then if you end up seeing bad performance over a remote network, then you must compare it to the baseline and that will tell you how much the network itself is slowing things down.
Continue reading...
Continue reading...