[omniORB] Re: Sanity Check on Large Data Transfers

[W. Eliot Kimber]

Very embarrasing, but I was testing with the wrong file. Doh! I was
using a 100K file, not a 1Meg file.

With a 1Meg file, the transfer times are, of course, much longer.

I thought my numbers were too good to be true.

For a 1Meg file, 33,000 nodes, it took about 60 seconds to transfer the
data. I didn't take the time to figure out what the actual volume of
data was, but I presume it was about 30-40 meg. Note that these graphs
include many redundant pointers to the nodes in the graph so the structs
include many more than just 33,000 separate node structs--there are also
structs for every reference to every node (and I can probably optimize
out a number of those in the transmitted struct, now that I think about
it).

Still, the performance for a 5000 node, 100K data set is pretty good and
is a more typical case than 33,000 nodes/1Meg original data.

Cheers,

E.
-- 
. . . . . . . . . . . . . . . . . . . . . . . .

W. Eliot Kimber | Lead Brain

1016 La Posada Dr. | Suite 240 | Austin TX  78752
    T 512.656.4139 |  F 512.419.1860 | eliot@isogen.com

w w w . d a t a c h a n n e l . c o m