Re: Understanding Terracotta caching
Date: Fri, 22 Jan 2010 09:20:47 -0700
Message-ID: <3c5f7821001220820v649e7541wd37d99207c653da2_at_mail.gmail.com>
Fallback plan? What's that? I swear some of these people don't have a clue when it comes to effectively coding for / managing a production application. I have been requesting fallback plans for every code promotion or system change for the past 3 1/2 years. No one seemed to think it was necessary until we broke the application in our last code promotion and they didn't have a backout plan in place. Took 6 hours to get a plan together and back out the bad code. Fortunately, we didn't have to pay any penalties to customers.
Back to caching -- I did make a little headway with the data architect. Following the suggestion to get cache hit ratios on the tables, I have convinced the DA to take another look at the tables he wants to cache. He still wants to cache the m table but I'm not sure it's really a good idea. For every logical read, there are 226 writes. That may or may not be a bad thing. He did mention that most of the tables he wants to cache would probably be refreshed once every 24 hours. I'm not certain that's soon enough for some of the tables he has in mind given the number of writes in a 24 hour period. I need to know more about how this is going to work. I also mentioned to him that I'd been reading up on Terracotta on their website to determine the best objects for caching. He invited me to attend the Terracotta training next week. I figure the more I learn the better prepared I'll be so I accepted.
--
Sandy
Transzap, Inc.
--
http://www.freelists.org/webpage/oracle-l
Received on Fri Jan 22 2010 - 10:20:47 CST