Hello all, this is a long post so bear with me....

I'm trying to troubleshoot some performance issues I've been experiencing on
a few of my servers, here's a few details and some specifics on what I've

2 servers in particular, both identically configured Dell 2650's with 2gig
ram, dual 2.4ghz CPU's, raid5 (3x36 drives), single 100mbit eth card at
100/full (switch settings match). Both servers have over 20gb free
diskspace on the / partition.

OES, edir, and IDM2 doing some NDS to NDS synching

Ok, here's the situation, these 2 servers, I'll call them lsr1 and lsr3 are
in the tree at the root level and hold 70 real replicas on each (I have
other root servers as well, more on this later), these servers do nothing
but replication (no user access/file/print etc...)

The dib size of lsr3 is about 550mb and 250mb for lsr1. I've spent a fair
amount of time tweaking the database cache settings and have them both
configured with hard limits and average from mid to upper 90% range for
requests serviced from cache.

Now, the problem I'm seeing is what I feel is WAY too much skulking
(especially to lsr3), I'm seeing last successful sync times between 30 min
to almost 2 hours (got some really slow wan links that may attribute to
some of this)

Notwithstanding, health is in good shape, but things are just slow. For
example, a local repair on lsr3 may take a couple hours to complete.

Is it possible that I've just 'hit the wall' as far as my hardware goes ? or
are there some other performance tweaks that I could try to help improve
things a bit ?

Now, a side note and question... I also have several root level
SLES8/ boxes (on identical Dell 2650's) that are also holding
about the same amount of real replicas with dib set's at about 250-300mb
each, these servers are running nicely. Now, the thing I've noticed here is
when I look at the ndsd pid's on sles8 (ps -ef | grep ndsd) I see about
90'ish pid's or more (1 for each replica ???) when I look at this
information on ANY OES server I have I only see 1 pid, is this normal for ? or could this be part of my performance problem ?

Any suggestions would be greatly appreciated