7189 - Speed up RDF initialization

Status: VERIFIED INVALID

(Core Graveyard :: RDF, defect, P3)

Description

[leger]

TO DO for Performance:
When the code to initialize RDF gets hit.  It seems to take quite a
  while.  Is there anything we can do to speed up RDF initialization?

Comment 1

[Chris Waterson]

How do we know that this is the case? Is there a Quantify trace around that
indicates this?

Comment 2

[leger]

jevering or phillip, can we get a Quantify trace please?

Comment 3

[phillip]

i'd give you one if i knew how to run quantify. it's on my list of things to
figure out... (is there some tutorial out there i can look at?)

Comment 4

[Chris Waterson]

Moving to M8, when DP and I plan to do some focused startup performance
analysis.

Comment 5

[Chris Waterson]

Clobbering this bug. No specifics. Not useful.

Comment 6

[leger]

Let's not "bag the bug" because we don't have enough for you yet.

waterson, I'll assign to me while I get you some quantify data, then throw back
over to you...

Comment 7

[leger]

Clearing Invalid...phillip come see me regarding quantify

Comment 8

[Chris Waterson]

I think it _is_ bogus to re-open this as a bug. We have _no evidence_ that "RDF
initialization" is a bottleneck. How do I know that? Well, there is no such
thing as "RDF initialization". So this is sort of like sending somebody out to
hunt for snipe. :)

Comment 9

[leger]

This bug was given to me by jevering...jevering, do you want to mark this
Invalid?

Comment 10

[chris hofmann]

waterson,  if RDF doesn't get initialized how does it run? ;-)
  -snipe hunter's anonymous

Comment 11

[Chris Waterson]

My point is, there is no "globally initialize RDF now" routine that gets
called. Each _individual_ datasource does initialization when it gets created.
The RDF service creates some hashtables when it's created (but that's about
it).

Comment 12

[chris hofmann]

moving over to dp.  if dp can find something here lets reassign
to waterson, otherwise lets close this.

Comment 13

[James Everingham]

We have done sufficient performance analysis to rule this one out now.  The
problems are more registry related at this point.. closing out.  This may have
been filed without fully understanding the problem (yes, from my report).

Comment 14

[phillip]

marking verified per jevering's comments.