HowTo monitor multiple hosts Applications

[simonquentel]

I want to achieve service oriented monitoring for my company.

I have troubles conceiving this monitoring, mainly because Zabbix is host-oriented.

Let's use an example. Let say that my application is using:
4 databases on 4 different servers (server_A,server_B,server_C,server_D)
filesystems on 3 other servers (server_E,server_F,server_G)
a web client at URL http://XXXXX
a windows service on server_H
A Linux process on server_I

I would like to monitor in real time the availability of the application as the global availability of all the previous Items.

In a separate view, i also want top monitor the availbility of all servers (regardless of the applicatio the are linked to).

Is there a means to do so without having several templates on the servers? I have the feeling that I should have one template for monitoring Server_A (CPU, ping, Disks, Filesystem) and another template with only FS of my application.

More generally, How can I configure criticalness propagation so I can see in real time the availability of the application.

I guess my explaination might be hard to understand so I can try to refrase if needed.

Thanks in advance.

[simonquentel]

SLA Calculation

Linked to the previous post, I also have a question regarding maps and SLA calculation.

As far as I investigated, the most flexible way to describe an application availability is to use maps and navigate from a first application availability map toward more technical ones including the technical root causes for these unavailabilities.

Is there a means to calculate the SLA of a map icon? It seems that SLA calculation from the IT services is way less flexible and directly linked to hosts (which again is to restrictive from an application point of view)?

Thanks in advance.

[Colttt]

you can create an Group with name: MY_APP

and in this group you can add all you servers, and then you will see if you app is ok or not..

i hope it will help you?!

[simonquentel]

Thanks for your reply.

The problem is that doing as you suggest means that an application / service is a group of host, when actually it's not that easy!

Again, let's say I am controlling Server_A for infrastructure monitoring purposes. I will for instance monitor CPU, disks, virtual and physical memory.

But for the application I might be intersted only in pinging the server (just to confirm it is up) and some disks ( where the application has stored data)

Just adding the host to the group wouldn't be efficient! Or am i missing something?

One solution I saw here and there is to create a host gathering the items I am intersted in, but in that case it turns really unscalable.

Any idea?

Thanks,

Simon

[simonquentel]

Am I the only one facing this kind of problems?

I am sincerly doubting that Zabbix can provide application oriented monitoring, that it is dedicated to infrastructure monitoring due to its architecture.

Is there someone that can proove me wrong? Any example is welcome.

[bschmidt001]

IT Services

Try looking at IT Services in version 2.0.x and later. It allows you to group items together to consider them as a "service."

[simonquentel]

Yes it does but in a non-scalable way.

Let's say I have 500 dbs and want to monitor the availability of Oracle as an application, I would like to check the trigger regardless of the host. I won't add manually all the 500 items!

Simon