We are running a demo of this. It started out as an eyeballing a netflow
collector and analyzer I dont have to poke all the time. we started
scrutinizer, liked it, but found out the price scale killed any chance of
getting it approved

the pricing for this wasnt as bad, and the sales guy has some incentives,
but the whole package was alot, and I didnt intend on even looking at the
monitoring side because port based pricing models can quickly get out of
hand

as part of the initial configuration i seeded the auto discovery just to
get through the setup. in the mean time, some other stuff came up and i i
got busy, this was friday or thursday

we have been having some intermittent issues with periodic slowness to some
customers, the symptoms were that of a bottleneck. We had to throw some
static routes into our OSPF network defeating dynamics to force traffic out
one connection, thinking maybe it was a saturated lower quality upstream,
no noteable relief. so we thought maybe we were saturating a backhaul that
was getting to high percentage utilization, we added a redundancy and
further split traffic up with static routes. no joy. it was at a point
where the next step was just going site by site auditing every device...fun
since the issue was intermittent, that means multiple times

the sales guy wanted me to commit to getting this thing up and running by
this weekend so next week we could list out what we want from it and how we
achieve it, or if we cant do it.

so yesterday i go to turn on the flows and send them to the server, the
weird slowness is going on so its irritating me.

i decided to clear out the alarms from installation and low and behold
theres an alarm on a named interface of one of the routers i tossed in on
discovery saying 90 percent or more usage. this is a 366mb licensed link on
a gigabit interface, so im quite curious. I drill into the detail, the port
is running at 100mb and saturating, i flap the port and its back to gigabit.

we only monitor with powercode currently, we have snmpc but its old and
shut off. Ive toyed with a whole bunch of other opensource and low cost
systems but never had enough time to actually drill down and learn them, i
did just get a book on nagios because it was cheap on ebay.

powercode is worthless for any amount of invasive alerting or monitoring at
any detail, if i want ports identified other than by port number it
requires an individual probe. pita. its good for long term static
monitoring and some real time tools, but its not an NMS.

the point here, is the solarwinds tool is sweet, and for the 100 interface
package with a promotion the cost is doable if one takes into account the
time investment of the other opensource platforms, installation, learning
curve, back end configuration, and plethora of gotchas.

this particular issue could have cost us a good deal in man hours tracing
it, refunds to customers for service impacts, and potential long term loss
of customers.

now, once i knew where the issue was, i knew exactly where to look in our
existing data to verify it. 20/20 hindsight doesnt mean those are the
toolsets that would have been picked out first. if this tool had been in
production use, we would have known the first time the link negotiated
down, and addressed it before there was any noteable service impact.


If you are very frugal in your interface selection, this can be a good
choice for an nms (i havent played with the atlas map other than dropping
some stuff on it) if you dont want to dick around with a diy solution. its
cheaper if you dont add the netflow analyzer package. Its solar winds so
its pretty, and user friendly. the flow analyzer does route monitoring too,
i havent looked at that, but the salesguy says he thinks we can visualize
our ospf with the network atlas component, if thats the case the boss will
likely drop cash. licensing is perpetual with 20% yearly for maintenance if
you want it



http://www.solarwinds.com/network-bandwidth-analyzer-pack


-- 
If you only see yourself as part of the team but you don't see your team as
part of yourself you have already failed as part of the team.

Reply via email to