RE: Selective data replication across WANs

One thought would be to replicate on a batch basis. Rather than having a server, which reads the changed records and then does a remote insert, delete, or update as appropriate, have the server read the changed records and output them to a file suitable for load or dbload. The server may then ftp the file to the distant location where its reception is acknowledged. The receiving server can load the records into a newly created (temporarily used) table from which you may insert, delete or update records as appropriate. The file being transmitted will require information for each record to indicate the appropriate way to process each record or you could use separate files for each process action; insert, delete, and update. Chuck Francis AT&T SPA All opinions are mine. Implement any way you like at your known risk. ---------- From: ilist To: internet!rmy.emory.edu!informix-list Subject: Selective data replication across WANs Date: Thursday, August 10, 1995 2:54AM Has anyone had any experience in replicating data across WANs using Informix? We are attempting to replicate records selectively by table from a database in Auckland New Zealand to a system in Antwerp Belgium. In the absence of any log sniffing software from informix or anyone else we have built a server which reads the changed records and then does a remote insert, delete, or update as appropriate. The problem is performance. Running across a tcp-ip link between two machines in the same room we are only writing 2 records per second. Informix advises us that we are correct in our assumption that a connection is being opened and closed for each line being written. They say there is no clever way to get around this. We are presently looking at the possibility of dropping flat files out and re-loading them at the other end. (Yuk) Any help very much appreciated Peter Tashkoff NZ Kiwifruit Marketing Board Auckland, New Zealand (ptash@iconz.co.nz)