database - Profiling Mnesia Queries -

Our Snine DB is running slowly and we think it should be somewhat faster.

So we need to profile it and complete it, what is happening.

There are several options which suggest themselves:

• run fprof and
• run the CPR and see which function is called a lot

Though these are both standard performance monitoring style tools. The question is how do I actually do query profiling - which questions are taking the longest time if we were an oracle or mysqul shop, then we would run only one query profiler, which will be used to run long-term queries Will return.

This is not a device that is available for Manasia.

The question then is:

• What is a profile present? Profile is available for mania - nobody thinks me, but prove me wrong :)
• How did you profile your questions and optimize your Manisa database installation

discussion

One of the problems with fprof as a profiling tool is that it only tells you about a particular query if fprof tells me that the X is slow and I Down to speed Tweak Again, less and see, Operation Y (which was fast enough) now the dog is slow, so I make a profile of Y and I realize that the way to make y accelerate is to make x slow so I am one of the bilateral trade-offs I'm ending the series ...

What exactly do I need to do? Stronger need is a way to manage multilateral trade-offs. I have a log of 2 metric shade-load real user activities so far that I can run again. What do I want to customize these logs?

  SELECT [fieldet] FROM [/ s]    
 Table] WHERE {field = * parameter *}, {field = * parameter *}  

and took 0.37ms average to run 285 queries of this form < / P>

They answer magic when it says: took 6.34s to run 17 questions on this form and scanned a full table on the table, you can Let me give an indicator on Land Y

when I have such a result on a representative set of user-activities, then I can begin to argue about trade-off in its round I- and design a test pattern.

The test pattern will be something like this:

• The activity will make X queries A, C and C faster but E and F slow
• test and measure
• then accept / reject

I am using Aralong which is 'know' that there is no question analyzer in this way, what would I want to know How is it that the other people (who have this problem) 'cause' about the Malaysian adaptation?

I have been suspended because I do not know more about Erlang or Manisa, There is a lot to know about tuning, and so far it seems very special from the discussion.

They get their fundamental point of view from most devices like Sound gprof such as tool, code function, i.e. functioning functions, calculation of exchanges, programs Counter sampling etc. Some people take a long time to practice that. Your disappointment device is normal for users like this.

There is a method which is less-known that you can consider, it is based on taking a small number (10-20) samples of program status at random times, and every one To understand, usually instead of summary, it means checking the call stack, but you can also check other information. There are different ways to do this, but I use the Stop button in a debugger. I'm not trying to calculate the exact time or invitation. These are indirect clues on best, instead I ask about each sample "What is this doing and why?" If I think that this is doing some special activity, such as performing X queries, where that object z is, and it is doing more than one sample, then this is the fraction of a sample that it There is a rough and reliable estimate of the fraction of that time. It is likely that this is something about which I can do something, and get good speed.