OpenNMS 101 - Module 0: Introduction to OpenNMS Video Lecture | A Complete Guide: OpenNMS - IT & Software

hi there my name is char spay log and
I'm one of the maintainer of the open MS
project now I've been told that you
young folks out there really prefer
watching videos to digging through reams
and reams of documentation online and so
we decided to put together a training
course we're calling open an MS 101 and
this is the very first lesson in our
open an MS 101 video series
it's called introduction to open an MS
and the idea is to give you a little bit
of the history of open an MS to talk
about the application a very high level
and to give you a glimpse of where we
hope to take the product in the future
now open AMS was started way back in
1999 we're not really sure when I've
been told July but I wasn't involved but
what we do know is at the very first
code with the open MS name was published
on SourceForge on March 30th of the
extent that was almost 17 years ago I
was brought on board in 2001 by becoming
korokke lab which was maintaining open
invested of time now oculus business
model was they had open an MS which they
used as a platform on which they built
the network monitoring appliance so they
put some extra software on top to open
an MS and they sold it as an appliance
to small and medium businesses I was
brought on board to build a business
around the open and ms part of the thing
namely a services and support business
now I started on September 10th 2001
which is very easy to remember because
in September 11th 2001 we sat around and
watch TV like the rest of the world so
it wasn't exactly the best time to start
a business but what we managed to do
okay now in May of 2002 couple things
happened first of all we released open
an immense window yeah we have one oh no
relief so that was kind of cool and the
second thing was that ocula and the
investors mainly of ocular and decided
that they were no longer interested in
publishing anything under the open
source model now they own the copyright
if you own the copyright you're free to
do what you want to
the code and so it was alright for them
to do that of course once something is
published under an open-source license
you can't unpublish it but all future
work was going to be proprietary and
they weren't going to release the code
to the public now I've been doing
network monitoring network management
since about nineteen eighty-eight and
and I saw the potential in open an MS a
lot of network management is this this
writing of glue where you take disparate
best-of-breed products and try to glue
them together into a solution and I said
hey if we've got this platform that's
open source then all of us management
geeks can get together and work on
building something in open in a mess
instead of having all this weird little
pieces of code that's not very well
maintained so I would notice those CEOs
office and I said give me open and mess
and they kind of looked at his watch and
he said well if you're off that payroll
by Friday we'll sprinkle water on you
and you can be the open house guy that
was the easy part
I don't know how many of you were
married but at the time I was pretty
much the main breadwinner so I went home
to inform my wife that I'd quit my job
and a starting loan company to sell free
software let's just say it didn't go
over that well I have since learned that
this is what I've been informed as a we
decision not necessarily a me decision
but we're still together so I guess we
lived through that and I kept the
project bowling until September of 2004
where I was able to convince a number of
other people to join with me and we
formed the open animus group and the
open NMS group is the commercial company
behind the opening this project and has
been since September of 2004 now when I
got started with my own business back in
2002 everyone said well you need an
elevator pitch and I'm like what's an
elevator pitch and they said you need to
be able to express in maybe 30 seconds
or so what your product is so if you
meet someone in an elevator and they say
what's open in a mess you have a certain
amount of time and so I came up with
this phrase open in a mess it's the
world's first enterprise grade network
management application platform
developed under the
poor schmuck ding doors open now the
problem is I'm a technical guy I'm not a
sales guy I can't close the door much
less a deal so I'd I hate marketing
snake I hate you know we are the market
reading paradigm changing synergy
building blah blah blah blah blah so
whenever I have a phrase like
world-first I want to explain it so all
of these phrases that we have
highlighted on this slide I'm going to
break down world's first probably the
grand old man of open source monitoring
tools is some cool Nagios it actually
started off a life as net st. and it
started out in the year 2000 January of
2000 so we registered sort in March of
2000 so we're contemporaries so met st.
ran into a trademark issue so they've
changed the name and that became Nadia's
there's a couple of other tools kind of
from that era that are around so we've
been around a lot so when a favor
world's first I basically mean we've
been around a while enterprise great now
a lot of open source tools start off
life because some developer wants to
scratch a particular itch and he writes
a quick piece of code to do that and
someone else finds it useful and they
start building upon that and then they
realize oh crap it doesn't scale so now
we need to rewrite it and rebuild it and
stuff like that open in a mess started
out as day one we wanted to run our tens
of thousands if not hundreds of
thousands of devices and our ultimate
goal is to basically be unlimited and so
we put that in mind we set out to build
open animus to do that which is one of
the reasons it's one of the few tools
that does scale now when you talk about
scalability it can mean many different
things to different people
and when I say scalability I'm talking
about well we can monitor discrete
devices we have customers monitoring
over 100,000 discrete devices with one
instance of open an MS it can also be
performance metrics we have other
customers who may not have as many
devices but they want to collect
millions of data points on bandwidth it
could be temperature it could be a ton
of different things and open an MS can
handle these millions of ultimately
billions of metrics we also handle
lot of events per second several
thousand events per second that we can
run through I think 30 or 40,000 is one
of the messaging systems we look at
recently and then remote monitors we
have this ability to deploy a monitor
remotely and we have one customer that
has a chain of retail stores and they
put an open in a mess monitoring each
one of those stores and there's over
4,000 so enterprise great now this next
phrase application platform this is
probably the hardest concept to get
across because when people think about
applications they might think about in a
world of warcraft so download world of
warcraft you start it up and boom you've
got your game you play it you don't
expect to do anything you're not going
to modify it you're just going to use it
you hope that they fix bugs but you're
kind of done well in the open source
world it's hard to build a business
around just an application that just
does one thing and you stick it out it's
done with opening a mess it's more of a
precise form yeah it does a lot of stuff
out of the box but it really shines when
you start to integrate it tons of things
big analogy I like to use is my wife is
an accountant and she basically manages
a two billion dollar budget with Excel
now granted she has tons of databases
and other tracking tools very expensive
tools that integrate with the
spreadsheet but at the moment she's
talking about using Excel to manage this
budget now if I went to Microsoft and
bought a copy of Excel and call them up
and said hey it's poor guy where's my
budget they laugh at me you know Excel
is a tool opening a message a tool and
to make it integrate with many things as
possible with terms of integration
points first of all we have a
full-featured rest interface relational
state transfer this is the ability to
send information with a with a call to a
URL either a get or put to a URL if you
go to either the Apple App Store or to
the Google Play Store you will see that
we have an opening a method called
compass that app which is visually
striking or very easy to use it's all no
phone rest there's no code outside is
there's special complex code in the open
in this platform itself compass just
uses restaurants
and we're continually improving the rest
interface so that other applications can
integrate with our platform built-in
open an MS is a strong event management
system and so we were able to use the
events and notifications to run external
commands to do even further integrations
and we have a bunch of built-in
integrations we have an integration with
rancid which is a configuration
management tools for Cisco and juniper
devices you can directly provision open
MS from your DNS over the Mexican check
your DNS to see if those changes update
itself and we have a very strong trouble
taking the API for such things as JIRA
and remedy and then finally this idea of
open source fully 100% of all of open
NMS is available under an open source
license
we don't do open core we don't do fopen
source there's not a piece that's extra
that you have to pay for and you're in
another proprietary license that you
have to to get all the cool features we
just don't do that now the main
application is published under the faroe
GPL version 3 but we do have pieces of
it like our integration with newts which
I'll talk about towards the end of the
presentation that it's published under a
more permissive license such as the
Apache life now if you type in open
source network management into Google
we're usually the number one hit I think
the same goes for open source network
monitoring so after the ads you'll see
right there that it says open an MS
after after the ad we don't do anything
for that we've got a very diehard small
but active community in just four years
we've been the number one hit for that
I'm pretty proud of that
honestly we'll lose it someday because
we're not really playing any SEO tricks
but I think just want to point that out
now what is network management now there
is a definition of network management
called F camps and fans for five main
areas divides this realm of network
management into fault management which
is you're monitoring what's going up and
down configuration accounting who's
using what on the network performance
measurement you know what's going
through my network what
River router and things like that
insecurity and open MS is really
strongly the fault management in the
performance measurement pieces we do
some configuration some security and a
little bit of accounting
now there's another definition of
network management which is basically
anything on the other side of the
keyboard which is kind of have my father
abused the network you'll call me up and
go hey the network stack and I know how
do you know that networks bent and he
says well I can't get Emily email and
I'll say what can you get to Google you
know me going yeah I'm get to Google
then it's not it's not the network give
some other issue which is why with
opening a mess it says network
management platform but it's the whole
stack from physical layer up through
applications and the tool can monitor
pretty much all parts of that but that's
why we call it network management system
because that's just a historic thing but
if in a way my dad thinks through the
network not necessarily application
management versus middleware versus
database management etc now there are
four main areas of open a mess each one
of these has a ton of nuances etc but we
can divide open a dimension of four main
areas the first one is event open M s is
event-driven I have an architecture
slide at the end of this presentation
which will kind of show you the heart of
open M s is a process it deals as events
so we can generate our own internal
events we can receive events from
external sources we can modify those
events change them create notices have
actions execute generate alarms just
tons of stuff with events and we're
going to spend the first several modules
in this class talking about events the
next thing which a lot of tools tend to
skip is this idea of provisioning when I
say provisioning I mean provisioning the
open and immense system if I'm
monitoring several hundred thousand
devices there's no way that as automated
discovery thing is going to work I mean
you can't just ping a huge class b
network network ranges you most likely
have to have some other inventory system
that's external to open in a mess that's
doing all this and then using our REST
API etc you can provision open and amass
to monitor exactly what's important and
ignore what isn't now back in the day
back in the late 90s early 2000s when
openness was started the big buzzword
was SLA service level agreements which
was how available is your your network
and there really wasn't anything at the
time that would do this I mean if you
had HP openview network manager you can
generate some estimate P values and you
could ping us a web server and you can
say well okay I'm pinging the web server
and it's up and I'm generating some
pages so I assume it's working
that's kind of like trying to measure
the fuel economy in a car and saying
well the fuel pumps rotating at such rpm
and the exhaust manifolds at this
temperature etc we've is decided to do
synthetic transactions let's drive the
car certain amount of distance and see
how much fuel we used and so a big part
of a venomous is this I be of service
monitoring of going out on the network
finding a network resource and making
sure it's responding in a proper manner
and in a timely fashion and then finally
performance data collection this is a
biggie let's collect you know bandwidth
statistics temperature etc all this
stuff on the network now there are two
main open intimates versions the one you
guys should probably play with first is
called horizon now horizon is kind of
analogous to the Fedora project it is a
version that we come out with very
frequently we have a major release about
every four months all the new cool stuff
goes into arising one thing we trigger
whether they open the next group open
and they're open an earnest group is is
the company behind open in this and we
have a bizarre business model we call it
spend less money than you earned
we've never raised a bunch of
investments we've never you know gone
out and got BC money on so we basically
survived all these years by making money
we don't make a lot of money because we
plow all of our profits back into the
product but we survived by making money
and so we've been trying to adopt best
practices and there's only one company
that seems to have done a really good
job in this
open-source space and that is red hat
and red hat adopted this model of
basically a subscription where you get
an ultra stable version of their main
distribution and so we adopted that
model with an open event so we're Rison
is like fedora all the new cool stuffs
in there and all the great new features
and stuff go into arising now lagging
horizon is something we call meridian
and the readiness of ultra tested very
very stable version of open in a minute
now the main difference between a rise
in meridian is that we charge for access
to read just like Red Hat so you buy a
subscription you get access to the
source code and the RPMs now you can
choose the same license that horizon is
published under which is the a GPL
version 3 but we also have a proprietary
license which is very very similar today
to tell certain customers of ours are
very large and they don't allow any a
GPL software in their organizations so
we have a license that they're they're
much more happy with but you do have a
choice every bit of code is still
published under an open-source license
now horizon as I said is released about
every three to four months we do a major
release from reading every year and the
idea is that if you're on a particular
version of Meridian that will do
security updates etc for three years so
if you're running meridian 2015 which is
last year's version or two years ago
it's now 20 something that will be
supported until 20 2018 comes out so
this year we would be releasing 2017 in
the current release of meridian is 2016
now here is this is totally extra credit
this is a general slide talking about
the architecture of open MS now very
center of this you'll see this process
called event D open MS is a Java app and
all these sub processes we call daemons
and so event D is the event daemon and
it is a publish and subscribe messaging
bus so basically processes can send
events in to emit D and then they can
subscribe to them they can say hey when
something happens let me know we have a
set of demons for monitoring
d-does our service polling collect
d-does our data collection we have a
provisioning system called provision D
external events we can get FTP traps
with units v logs we can get xml
messages there's even a transactional
language one daemon which is for
telephony interfaces automations we have
an alarm subsystem we have notifications
we have trouble ticketing since
basically hey you want a particular
event happened let me tell somebody you
know if it's particularly it meets a
certain set of criteria you can send it
out there we also have the open in this
web app which we're going to spend a lot
of time in but because of that REST API
that I was telling you about
we also have this ability of timing to
the system externally do checkout open
in this compass it's a great little app
and again it is totally done through the
REST API now open MS stores data in a
couple of places we have configuration
data which is just in files that are
stored in filesystem we store a lot of
information in a database currently
we're limited to PostgreSQL now we'll
stay limited I mean per square scale is
a great little database but our goal is
to make open an immense database
agnostic we're not quite there yet and
for our data storage we can store it in
round-robin database format which is
similar we actually use rrdtool so if
you've ever used cacti or anything like
that it's very very similar to that but
in situations where it will scale we
also have a system called newts newts is
built upon a Apache Cassandra and it's a
new time series database that is
incredibly scalable and replicable et
cetera which kind of brings me to where
we want to take open an MS a lot of
people have talked about this thing
called the Internet of Things the IOT
the Internet of Things where you know
we're talking billions of millions of
devices and billions and metrics and
that's kind of where we're targeting
open in MS right now we'll do a lot of
stuff to position opening this to be a
platform for the Internet of Things
loots is a key part of it here's a graph
this was at a customer up in Maryland as
you can see here we were recording about
51,000
as per second one of our guys is getting
ready to do a presentation where we can
do a million events per second it
depends very much on how much hardware
you invest on the newts back in but
because of the nature of newts and it's
the underlying Cassandra database you
have a lot of a lot of scalability built
in there but what about gathering the
information I mean here we have you know
all these events per second but we
really need what if we don't have stuff
that's geographically diverse etc we're
now working on something called the
minion that's our kind of working title
for it and it's a small piece of code
that can live out on a remote box may be
a Raspberry Pi may be something like an
Intel nook but that is it'll do local
data collection and monitoring and then
send those just the events you're
interested in back into open an MS and
notice that it'll work behind firewalls
it'll work if it's directly connected to
the Internet cetera
and the idea is that no matter how big a
network you can have thousand reasons
let's say you manage a hotel chain where
you could put a minion in each hotel or
a minion on each floor and it can
monitor your door locks and your minibar
and the internet connections in your
set-top box et cetera and this is
ongoing work right now a lot of it's
coming out in horizon 19 which will be
released in the next couple of weeks
so that's kind of it for open MS that's
kind of the overall view as I said we're
an application platform and so it really
starts to shine when you customize it
and we're going to spend a lot of the
next modules going over how you install
open in this and how you customize it
again this is 100% open source we hope
you find it useful and see you next time