IOUG-A Live! 2002 - Day 3 (Tuesday)
by Stephen Andert
He cited a couple of benchmarks that show HP is faster (at query processing) than Sun & IBM. Their tests show Oracle 9i RAC on HP achieves near linear scalability.
He had a demonstration showing HP Service Guard with TAF (Transparent Application Failover) as well as one without TAF. They appeared to work as expected and TAF looks like a valuable feature that could help provide the high-availability that many businesses require.
Rene Bonvanie was the other General Session presenter today. He is the Vice President of 9iAS and Oracle Technology Network (OTN) at Oracle. He wants to know how to improve OTN. He asked for feedback at firstname.lastname@example.org. If you haven't been to otn (otn.oracle.com), go take a look.
Rene stated that their testing shows that 9iAS is faster than .NET which would it make it a good choice for any Application Server. He did a demonstration on making db objects available to a portal. Also, he showed how easy it is to convert PL/SQL packages to a service that can be used on a portal.
As far as today's educational sessions, I again wish I could be in several places at once. During one time slot, I elected to attend session 515 which was presented by Craig Shallahamer. This presentation called Oracle Response Time Analysis was outstanding and complemented other sessions that focus on Wait Events. Craig defines Response Time as the sum of Wait Time and Service Time. Focusing exclusively on Wait Events might cause you to overlook a Service Time that is out-of-line indicating a CPU-related problem. Craig showed how looking at certain events you can easily track down the root cause of a problem.
I talked to somone in one of the other sessions that I wanted to go to (Michael Abbey's presentation Stand By Your Standby) and heard that it was good as well. I'll have to review the whitepaper on that one when I get home.
The quick tip Q44 by Bill Schott called 4 Steps to a Healthy Standby Database was another great session. Bill has been working with standby databases for a couple years and the school of hard knocks has taught him quite a few lessons which he shared with us. One of the things he mentioned was that by monitoring the unrecoverable_time from v$datafile, you can detect a problem that could prevent you from accessing your standby database if needed in an emergency.
He also provided a step-by-step process for when you need to add a datafile to the primary. One trick is to create the file small since you need to copy it over to the standby. Also, a file resize operation is handled automatically by the standby, so that process won't require any additional manual intervention to maintain the integrity of the standby database.
There was also a presentation called bulding a Simple SQL Interface to any LDAP Server. Stephane Faroult presented this and gave a good overview on accessing any LDAP server (not just OID). I also picked his brain to get some thoughts on some question I am running into on planning a conversion from tnsnames.ora files to LDAP.
The Big Bash is always a popular event. This year, the event was held at SeaWorld. in spite of some chilly temperatures, there were a fair number of brave souls that rode ShipWreck Rapids and a large number of people sat in the "Soak Zone" for Shamu's House of Douse. I stayed dry other than putting my hands in the dolphin tank to pet them.
Tomorrow looks like another full day of running from session to session and trying to get a few more minutes on the exhibit floor to get some additional information and maybe spend a few minutes in the bookstore.