[req] Migrating from HP-UX to Linux?

Hi to all, my company is considering to port our Informational System from Informix Online for HP-UX to Informix Dynamic Server for Linux. I'd like very much to have your opinions and advices (or maybe some link to articles and road-test). We have both local programs connecting through ESQLC (which are written by us so we can recompile and even modify the code) and remote application connecting through ODBC. In particular, 1- how much similar are installation and manegement? can we assume we can reuse the know-how we acquired on HP-UX version? 2- how similar is the SQL implementation? 3- does the ESQLC code need to be modified? 4- anyone has ever tried to run IDS on a cluster of machines? with what results? 5- how IDS for Linux perform and how much reliable is at enterprise-level burden? Thanks for any help and best regards, -- Cristiano G. Gozzini Sent via Deja.com http://www.deja.com/ Before you buy.

cgozzinirobur@my-deja.com wrote: > > Hi to all, > my company is considering to port our Informational System from > Informix Online for HP-UX to Informix Dynamic Server for Linux. > > I'd like very much to have your opinions and advices (or maybe some > link to articles and road-test). > > We have both local programs connecting through ESQLC (which > are written by us so we can recompile and even modify the code) > and remote application connecting through ODBC. > > In particular, > 1- how much similar are installation and manegement? can we assume > we can reuse the know-how we acquired on HP-UX version? Except for OS related differences in kernel configuration and monitoring tools, identical. > 2- how similar is the SQL implementation? Between HP & Linux, identical. Between OL5.xx and IDS 7/9.xx? There have been a few enhancements but it is backward compatible. Only gotchas: o OL5 was pretty stupid so you will see different query paths in 7/9 o If you depended on 5's tendency to use an index for ORDERing without sorting to return first data for interactive apps quickly forget it. Not only is 7/9 smarter but it sorts several times faster so it tends to select indexes for filtering value instead and just sort. The new optimizer directive {FIRST_ROWS} or SQL command SET OPTIMIZATION FIRST_ROWS will cause the optimizer to favor query paths that sacrifice total cost and speed for the ability to return the first results quickly. o UPDATE STATISTICS; is not enough. Peruse the Performance Guide for the guidelines on using the new UPDATE STATISTICS options efficiently for best results. Or just get my dostats utility from the IIUG Software Repository in the package utils2_ak. o Watch for correllated subqueries. IDS will unwind them into simple joins internally which will be faster IFF you have the indexes to support the join. Else it will be slow so you may have to review your schema. o There are many new features to take advantage of but none are required. > 3- does the ESQLC code need to be modified? No, except to take advantage of new features like multiple connections and threads. > 4- anyone has ever tried to run IDS on a cluster of machines? > with what results? IDS 7/9 will run fine on a cluster as long as you do not try to share drives that they access. You can get that functionality by using IDS/XA 8.3x. > 5- how IDS for Linux perform and how much reliable is at > enterprise-level burden? Like a rocket and solid as a rock. > Thanks for any help and best regards, Art S. Kagel