Oracle Maximum Availability Architecture(MAA) is a set of high availability best practices which targets to minimize the complexity in designing the optimal high availability for critical production databases. I do not work as an infrastructure(operational) dba for years. Within last years especially with 10g Release 2 Oracle’s availability options evolved a lot. I try to catch up with these on my laptop, thanks to vmware, and I really admire how everything is as is now with Oracle.
When I think about the recent years I may easily give my job description like this; database developer, part-time fire-fighter and part-time myth-buster. I always loved to develop database applications, I am working with application developers as an old Oracle dba for the last 4 years. Mostly database modeling, physical design, Oracle’s optimizer and testing for performance is what I think I am useful. Also if there is an Apex project I never miss the chance, developing Apex applications is always fun for me.
I used myth-buster on purpose, because lots of my time is dedicated to exorcism of some evil believes which mostly depend on rumors or some n years aged wrong and/or undocumented test results; “never use Oracle xyz feature it is buggy” or “this optimizer you call in fact is an abuser, I never get what I need without hints” kind of very common behavior. Usually it feels like being Van Helsing; read the documentation which is available to anyone and easy to understand, not rocket science at the end, and kill the beast with a small test case which is most probably adapted from Asktom or the Oracle Users’ Co-operative FAQ kind of resource. In the beginning this is a delicious rant but after years it becomes boring, disappointing and dissatisfying. What is the rationality behind not reading and trying to understand the technology that you are using to make money, please advice me? These days everybody is after some quick and dirty solutions.
I used fire-fighter on purpose, because I am called for aid when something really goes crazy or unexpected on production after a long development and testing period. These cases are the worst ones since with my experience it is usually too late and changes needed will cause delays and angry customers. Still it is a fire and I do whatever I can, in most cases interestingly even this amount is enough. But this is Oracle’s success of course, there is always some options guide you directly to the root cause of the problem and then a workaround related. Like Apex I am also a fan of 10g Grid Control, I never miss the chance to use it when there is a performance problem especially.
During Mr.Alejandro Vargas three days seminar I strongly felt that I have to leave this mystique world of mine and return back to the world where checklists, metalink notes and documents rule. If you do not follow them you even do not get to the point where you ruin everything. How unbelievable it is that anybody can develop some piece of code, google and copy&paste are the major enabler technologies here. When it comes to Oracle database, how ironic also that it works very satisfactory with its defaults and let people think they are “developer”s or “dba”s. I believe even after 10g’s self management capabilities and Grid Control heaven, today database market need experienced people more than ever. Do you think anybody can implement RAC or Data Guard options without reading and testing carefully? And I seriously bet that if all Oracle users as a prerequisite only read the concept guide all around the world, if you consider Oracle’s dominant leadership in the market, we should easily expect a technological shift, of course my share should be less otn forums type of questions, fires and myths.
When I contacted Alejandro with the content of this MAA seminar my primary focus was to motivate our senior dba staff in using these features. We have unlimited kind of Oracle licensing, so avoiding these options we already purchased is both a threat and causing inefficiency for the company. Alejandro was the right person and luckily he was very interested with the content and as a result for the last three days was successful. But this was only the beginning, there is a big challenge waiting. Any expert who is motivated to implement these new MAA features with 10g Release 2 will be facing another very common behavior; “if it works never take a risk and change it” Of course managers need to avoid risks, but if this is the only priority you kill the creativity in your organization, how may a dba can take risk for the sake of performance of a system if you do not support him as his manager? The answer is simple, no dba would do that and you live with a release xx years old forever. So I plan the next step as to implement and document a MAA solution’s success, this will be a reference point I hope. This is my new priority and motivation for the coming months, so I really suspect that I will be ending up as being an operational Oracle dba again, but if something unexpected happens my b plan is already in pending status :)
I hope you enjoy my three days seminar notes below, much more is already published at Alejandro’s blog, he also may be publishing a post related to this seminar with our questions’ answers and his İstanbul experiences I guess.