My Configs and Projects on Metalink
I was asked to look at a
customer problem recently. I made the following observations and
I wondered if any other customers had any experiences worth sharing.
1) The customer won't give Oracle support an RDA for assessment because it is too invasive. This is the first TAR I have ever seen where an RDA has been refused, admittedly I am not in support, but I have worked with commerical and public sector customers and none objected to this collection of data. The RDA certainly gives support more data for their problem solving.
2) The TAR system itself seems to lend itself to describing an immediate problem, quite often it can appear that the customer didn't test the upgrade or have a backup plan or have an old system they could revert to. In reality this would not be the case, so perhaps it is due to the method of communication. Once the initial panic subsides, then the communication does improve.
In metalink we have tried to address this. There is a tab that allows you to set up My Configs and Projects. I think this is a great innovation.
3) Looking at the stats pack, there seems to be a lot of problem SQL. This seems a surprise to me. It is as if there has been user testing but no volume or stress testing. Is it possible that the regression tests are the wrong ones? I would be interested in any comments on good testing strategies, especially for bespoke applications.
It is intriguing to look at these problems with a dispassionate eye. There is rarely a silver bullet for these problems, I guess it is just a case of solving the immediate micro-problem whilst also looking at any grander lessons learned to try to stabilise the application.
1) The customer won't give Oracle support an RDA for assessment because it is too invasive. This is the first TAR I have ever seen where an RDA has been refused, admittedly I am not in support, but I have worked with commerical and public sector customers and none objected to this collection of data. The RDA certainly gives support more data for their problem solving.
2) The TAR system itself seems to lend itself to describing an immediate problem, quite often it can appear that the customer didn't test the upgrade or have a backup plan or have an old system they could revert to. In reality this would not be the case, so perhaps it is due to the method of communication. Once the initial panic subsides, then the communication does improve.
In metalink we have tried to address this. There is a tab that allows you to set up My Configs and Projects. I think this is a great innovation.
3) Looking at the stats pack, there seems to be a lot of problem SQL. This seems a surprise to me. It is as if there has been user testing but no volume or stress testing. Is it possible that the regression tests are the wrong ones? I would be interested in any comments on good testing strategies, especially for bespoke applications.
It is intriguing to look at these problems with a dispassionate eye. There is rarely a silver bullet for these problems, I guess it is just a case of solving the immediate micro-problem whilst also looking at any grander lessons learned to try to stabilise the application.
