MAXDB Development

[Former Member]

Hi,

I am trying to change the code of maxdb in order to make it forward a copy of each SQL command it receives to another maxdb instance (in another machine).

I have 2 questions:

a) How do I enable trace messages?

b) Is there a single point where kernel processes SQL messages?

Thank you

[lbreddemann]

Hi Georgios,

this reads as if you either want to have some kind of shadow-database or some kind synchronization.

For the later one: use the synchronization manager and for the shadow-database log-shipping and recovery would be the right thing to do.

Anyhow...

a) how to turn on/off the trace options is described here:

<a href="http://service.sap.com/sap/support/notes/837385">#837385 - FAQ: MaxDB database trace (VTRACE)</a>

Unfortunately the traces are not too readable for people that do not actually develop MaxDB...

Perhabs running the SLOW-Kernel when doing the traces gives more insight to you.

b) sql-commands are (to my limited knowledge in this area) processed over several distinct layers (parsing, building up an internal representation of the command, rewrite it, put it into shared sql chache, optimize it and finally run it).

The question would be, at what state do you want to send the sql-command over to the other instance? From my perspective, you would have to send the whole client data package also to the second instance, since the checks on permissions etc. might lead to different results in the other instance.

Moreover, what do you do with the result sets that you generate in the second instance? Or do you plan to shuffle over just the sql commands that actually change data?

All in all: what is the requirement you try to fullfill with such a code change? Perhabs there are allready solutions for what you try to do !?

Best regards,

Lars

[Former Member]

Hi Lars and thanks for your reply.

The link you provided requires SAP Services Membership (which I don't have). Anyway, what I am trying to do is make maxdb a distributed database. Sending the SQL Query to another database (or databases) should be the first step. What I plan to do afterwards, is build some logic to decide what data to store at which database node. Probably, I will have to implement clustering to a lower level (perhaps at the page reading/flushing function)

Regards,

Giorgos

[lbreddemann]

Hi Georgios,

ok, this is a quite big venture you're going to start. Good luck with it!

Since you have no SAP account, you're basically bound to the information available on the internet and the SDN.

Nevertheless the MaxDB Wiki here in SDN is maintained quite good, so give it a try:

MaxDB Wiki ´

There you'll find the Support Guildeline and there is also a chapter about tracing available.

Hope that helps.

Best Regards,

Lars