Hello SQLiters! I am trying to figure out "best practices" for accessing SQL Tables from other servers in the same LAN with bandwidth of 1 GB. There are several purposes for such access: 1) to query tables from one of the servers running HTTP-Daemon (apache) using perl. 2) to read entire table once a day and summarize the information for trends. Files are of significant size and the summarized information is small. 3) to back up the entire tables on a daily basis. I figured that the solution to (1) is to use SQLRelay. I have never used it but found in the list of SQLite products. I am thinking of creating a Store Procedure for (2). Then, using SQLRelay, invoke it and save the results in the local server for historical trends. My thinking of (3) is along the lines of creating a backup file using the .dump command, followed by a remote copy (rsync, scp, rcp) to the server that handles the backups. We actually don't install SQLite in our servers and use it only within perl. So will have to figure out how to invoke .dump from perl. Background: We have 20+ Unix Servers (large SUN and AIX servers with 71 CPU's or so) and a Mainframe runing both zOS and zLinux in multiple partitions (LPAR's). These servers are all running perl daemons capturing performance information using "ps, vmstat, sar, and acctcom". Until recently I had been using csv to capture the information and then via NFS, I would read all the data into one single server for daily processing into a historical performance database. My HTTP server would via inet and sockets connect to the servers and pull data for daily monitoring. My backups are done on the MF using zOS/SMS/HSM for archival. This entire solution works but I am duplicating data across servers and we are talking BIG datasets. The nastiest problem with this approach: keeping records layout in sync across all servers. Then, I started using SQLite and realized that I could re-write all my daemons to create the Tables at source when the data is captured. This way, if the record layout changes, I would not have to modify anything else down the path but once at source. My next challenge is accessing the tables real time. I thought of writing my own inet/perl scripts to receive requests. I started checking around if anybody had done something along these lines and came across SQLRelay (http://sqlrelay.sourceforge.net/). Regards, Uriel_Carrasquilla
