Network Management

is there a guide shwoing the srver Specs needed to install Airwave Versus number of devices supported ?

Also is Airwave installed on Windows or Linux machine ?

 

The Airwave Server Sizing Guide:  http://support.arubanetworks.com/DOCUMENTATION/tabid/77/DMXModule/512/Default.aspx?EntryId=11469

 

Airwave runs on CentOS Linux and installs the OS as well as the Application on a bare server.

Airwave can be run on HyperV or ESX.  The install package INCLUDES CENTOS for the application therefore, it is not installed "on top of" another OS.  

 

The server specs are below:

 

 

 

Hi Telnet,

I invite you to read http://community.arubanetworks.com/t5/AirWave/Airwave-Disk-Space-Usage/td-p/19484 for some more context as well.

Yes the sizing guide is a good start but I have had terrible experiences with running Airwave on an ESX server and have moved to dedicated hardware and it has been light and day.


As for HD space, the sizing guide does not take into account the number of unique clients Airwave will gather information on and by not tweeking the historical data retention section of Airwave, you can quickly run out of diskspace if you are not careful.

Here is an example of a current customer of mine:

Customer with 2200 devices
AMP Server
----------
RAM: 64 Gb of RAM
HD: 1.5 Tb RAID-10 24x146Gb 15K (more than what the recommended was)
CPU: 2 x Intel(R) Xeon(R) CPU E5-2665 0 @ 2.40GHz Hyper-Threaded 8 Cores 20480 KB cache (2400.021 MHz actual)

Performance is great with this server after some tweaks because it is monitoring 21 active Aruba controllers and > 2200 APs but what happened to me is I quickly ran out of diskspace, yes even with 1.5Tb and this was due to me not tweeking the Historical Data Retention section. Our AMP server saw anywhere from 20,000 to 25,000 daily clients therefore...since every MAC address takes up a fixed size amount of RRD space which greatly depends on how long you have set HDR for different data collection points, I would take this into account when sizing a server.

The space hungry ones are 'Client Data Retention' and 'Client Association and VPN Session History'

It is very hard to size a server when you're using an Airwave to monitor hotspots like mine but if you are monitor a campus type location with repeat client then the sizing guide does an adequate job of sizing.

For example, a financial customer of mine with 1289 devices (APs and controllers), averaging out at 1,000 clients per day. I am only using 74Gb of HD space.

The key here is to make sure you have properly tweeked your server for the needs of your network.

I hope this gives a little more insight.

Thx

what is meant by IOs needed & IOPs needed mentioned @ the above table 

Disk reads and writes

The data maintained by AirWave is going to be composed of many read / write processes.  As the data is constantly updating, the IO requirement is very important.  If you have a medium or large deployment, you don't want to short the 15k RPM disks.  Some smaller customers have been able to use 10k RPM drives, but sometimes that comes with performance impact (they poll devices less often).

 

Also, if you choose the VM route, just be aware of the overhead.  I typically recommend scaling a server size higher if the deployment plan is on VM.  Also keep in mind to make sure that the AMP instance has some level of priority to the disks so that the IO speed doesn't drop.  If AMP is not able to read/write at an optimal speed, you will feel the performance (sluggish AMP UI, data not catching up, possible processes restarts, long waits for reports).

I realize this is a very old discussion, but it seems to still be relevant. I am new to Aruba and am trying to plan an Airwave installation starting with 3 APs, and gradually installing more. 

 

I am seeing that the Master console server requires a lot of IOPS. We are planning on using a VM for the server.

 

We do not currently have 15K drives available. Is it possible to use 10K drives? Will we see a performance hit? 

Are these Aruba Instant APs?  If not, then you're looking at the load of controller + APs.  Potentially switches too, if you want RAPIDS data.  While it will survive with 10k drives for a smaller deployment (like a lab), when you scale past 100 devices you may start to see the disks causing issues due to not keeping up.  You might also not be able to get the full advantage of AppRF with slower disks.