Tip:
Highlight text to annotate it
X
Understanding Orion NPM scalability and performance is really a key part of your
job as an Orion system administrator and a key part of your job as a network
manager. This training video will explain how to build, configure, and
optimize Orion NPM systems for scalability and performance. We will also
explain the requirements of the SCP program on this particular subject.
And within this video, we will cover things like Orion system settings,
licensing,
performance, and the Orion system architecture.
Let's go ahead now and take a look Orion, and we will delve a little bit deeper into
each of these subjects. When it comes to Orion NPM scalability and performance,
there are four main topics for us to discuss:
Orion licensing,
troubleshooting and diagnosing gaps in chart data,
the Orion additional polling engines and when to deploy them,
and when to use a centralized vs. a distributed network management system.
When it comes to Orion licensing, it is important to understand that Orion is
not licensed based upon the total number of monitored elements.
Instead, Orion is licensed based upon the highest number of nodes,
interfaces, or volumes.
So in the case of an Orion SL2000 license for instance,
you would be allowed to monitor up to 2,000 nodes,
2,000 interfaces, and 2,000 volumes,
for a total of 6,000 monitored elements.
Now remember, nodes are devices
like switches, firewalls, routers.
Interfaces include both physical interfaces like router ports,
as well as virtual interfaces such as subinterfaces and VLANs,
and volumes are simply logical disks as recognized by the system you are
monitoring.
Now that said, the element count is Important, because you want to think
about the total number of monitored elements
when planning for scalability requirements.
In other words, how much Orion can monitor before you need an additional polling engine
and also when thinking about the need for database storage.
Let's go ahead and take a look at Orion. If you go to the Orion home page
and click on Admin,
and then under Details click License Details,
you can actually see the number of network elements monitored,
in this case 1768,
and the numbers of nodes, interfaces, and volumes.
In this case, it would mean that you can monitor up to 1506
interfaces or you have monitored that many in the system,
and so you would need the Orion SL2000.
In almost all cases,
you can simply look at the number of interfaces, as it will almost always be
the highest number,
and therefore your Orion license will be limited based upon that number of
interfaces.
Understanding gaps and chart data and how to troubleshoot those is really
important,
because it can be a little bit confusing when you first look at a chart that has
gaps in it.
Let's think about how a chart is generated. When you request Orion to display a
chart, Orion then goes and pulls the dataset needed from the database
to build the chart,
puts the dots on the chart for the different data points
and then attempts to connect them with lines.
And so if you've requested a chart
with greater granularity
than the data in the dataset, you will start to see gaps or disconnected dots.
Now gaps in chart data can happen for several reasons.
First of all,
if you have simply requested
more detail than the system holds,
you will get gaps.
Secondly remember, if you have an outage or a change control window,
and the device you are trying to monitor or the interface you are trying to pull
is unavailable for part of the time
where you are trying to draw a chart,
you will see a gap for that time because there is no data.
Also, if you are overtaxing the Orion system
based upon either network latency, the performance of the polling engine,
or the SQL server we are trying to write data to,
then you might see a case where we are starting to run out of cycles
to process the data,
and you start to see sporadic gaps in the chart data.
Now, when you are seeing gaps in chart data,
you want to go ahead and take a look at the performance of the polling engine,
how often you are collecting data, and whether or not the engine is keeping up.
You will also want to look at the SQL server backend
to see if your disk
input and output read and write queues
are starting to grow
beyond the recommended limits,
because that could cause additional delay
and gaps in the chart data. When understanding the Orion additional polling engines,
there are a few important things to remember.
First of all, you want to understand the Orion system architecture,
and in a basic Orion system there are three main parts: the Orion website,
which displays the data in a Germane user interface,
the SQL server, which is the back end that stores the data,
and the polling engine.
Now, an Orion polling engine is actually a collection
of different processes and services that all work together
to basically collect the data and store it.
So a polling engine might include things like
listeners for syslog and traps,
pollers for SNMP data,
and additional pollers or services that collect data
like the universal device poller.
And so it is very important think about each of these components.
Now additional polling engines are available from SolarWinds.com,
and they are required whenever you are trying to poll more than one system can
handle.
Now a basic rule of thumb is that a single polling engine can monitor
about a thousand elements per minute,
or about eight to 10,000 elements
per system based upon the default polling intervals
which are nine minutes for data collection for statistics,
and two minutes for default status detection.
You want to also understand the requirements
that surround reachability in terms of additional polling engines.
So if you have a centralized Orion system
and from that central point you are trying to monitor two disparate networks that
have overlapping address space,
you will need an additional polling engine so that each polling engine can be
configured
to connect to the specific address space you want to poll.
You can't take one Windows-based Orion polling engine
and tell it to poll the same address space on two different networks.
It is simply a matter of reachability and routing, so keep that in mind. 0:05:56.869,0:05:59.510 If you decided to deploy additional polling engines,
you will manage those with this tool within Orion.
From the Start button, choose "Start",
"Programs",
"SolarWinds Orion",
and then under Advanced Features,
you will see the Monitor Polling Engines utility.
This application is very helpful, especially if you have multiple polling
engines,
because you will see those polling engines listed
here
and all connected to the same SQL server.
You can actually also use this application for several other things
including stopping and starting services,
and load balancing
your devices among your pollers.
It is also important to understand when to centralize Orion
vs. deploying a distributed Orion system.
A centralized Orion system will have these things in common:
first of all, there will be one SQL database that all the Orion polling
engines in the Orion server itself share.
Secondly, they will all be in one physical location or at least connected in a very
low latency network situation.
And typically, you will have one main Orion license like an Orion SLX,
and an additional polling engine for scalability, up to five in any one
single Orion system.
A distributed Orion system
would have multiple Orion licenses, meaning up to
actually
just more than two Orion SLXs,
and don't have their own separate databases,
and they will be separated geographically usually.
And in that case, you will use the Orion Enterprise Operations Console, or EOC,
to consolidate the views and data from these disparate Orion systems into one
common view.
Now there are many reasons why you might want to distribute your network management
system,
and we cover this in detail in the recorded webcast on distributed vs. centralized
network management systems, so you should watch that if you want to learn more
about this topic.
This concludes the video on Orion NPM scalability and performance.
Be sure to visit the SolarWinds.com website and click on education to learn more on this subject,
and as well as other subjects included within the SCP
program.