[igraph] Problem with very large graphs?

[Uri Shwed]

Dear Gabor et.al.,
I've been an avid user of igraph for a while, thanks for your great  
package!
I'm starting a new project with a very large data set (about 215K  
nodes, 3 million edges). I'm probably doing something wrong, but maybe  
I'm stepping beyond the capabilities of igraph or my machine  
(dedicated server with 16G ram)?

my data looks like this:

>CPP[1:10,]
   profile_id_a profile_id_b p_p_id owner_id                  ts active
1          2702        11073 635414        0 2008-07-24 03:14:28      1
2        156587        97680 635168        0 2008-07-24 01:45:35      1
3         17216       156593 633858        0 2008-07-22 02:27:47      1
4         99577       131264 686091        0 2008-09-26 18:47:23      1
5         89466        97680 635110        0 2008-07-24 01:45:34      1
6         57866       131501 619321        0 2008-06-26 01:49:51      1
7        156167          305 634744        0 2008-07-23 13:29:36      1
8         24475          305 634743        0 2008-07-23 13:29:36      1
9        156147          305 634742        0 2008-07-23 13:29:36      1
10       147075       149997 603846        0 2008-06-11 02:31:09      1

#In small quantities, it works:

> a<-CPP[1:10,]
> b<-graph.data.frame(a)
> b
Vertices: 17
Edges: 10
Directed: TRUE
Edges:

[0] '2702'   -> '11073'
[1] '156587' -> '97680'
[2] '17216'  -> '156593'
[3] '99577'  -> '131264'
[4] '89466'  -> '97680'
[5] '57866'  -> '131501'
[6] '156167' -> '305'
[7] '24475'  -> '305'
[8] '156147' -> '305'
[9] '147075' -> '149997'
> degree(b)
 [1] 1 1 1 1 1 1 1 1 1 1 1 2 1 1 1 3 1

#In large, it seems to work:
> GPP<-graph.data.frame(CPP)
>ecount(GPP)
[1] 2961141
#But it doesn't really:
> degree.distribution(GPP)
[1] 0.9999998
> unique(degree(GPP))
[1] 0
 Any ideas?
Thanks,
Uri