[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [hobbit] scalability of hobbit/xymon ?



Thanks for the proposition.
The hobbit clusters are in production but i think i can run some scripts on
a standby node. (morever we got preprod).

It'll be a pleasure to try your scripts and share results with the ML.

Regards,
Nico


On 08/07/09 mercredi 8  18:41, "T.J. Yang" <tj_yang (at) hotmail.com> wrote:

> 
> 
> To resolve this issue of unknown capacity, I think we need to come up a
> test/torture script for xymon hobbit server. This way we can
> demonstrate/simulate the load capacity of xymon server before deployment.
> 
> 
> I have some idea on how to do the coding(using perl's test module), it will be
> great if your company can allocate resource to do this programming task
> together. otherwise I am interested to do it myself but a lower priority.
> 
> 
> 
> 
> T.J. Yang
> 
> 
> 
> 
> 
> ________________________________
>> Date: Wed, 8 Jul 2009 15:00:26 +0100
>> From: tainted.soul69 (at) googlemail.com
>> To: hobbit (at) hswn.dk
>> Subject: Re: [hobbit] scalability of hobbit/xymon ?
>> 
>> Although I'm far from knowing what hobbit can do but doesn tthis paragraph
>> 
>> 
>> 
>> Quote : http://www.hswn.dk/hobbit/help/about.html
>> 
>> Xymon can handle monitoring lots of systems.
>> 
>> 
>> Big Brother is implemented mostly as shell-scripts, and performance suffers
>> badly from this. In large networks where you need to monitor hundreds or
>> thousands of hosts, processing of the data simply cannot keep up. Another
>> problem with BB is that it stores all status-information in individual files;
>> when you have lots of hosts and statuses, the amount of disk I/O triggered by
>> this severely limits how many systems you can monitor with one BB server.
>> 
>> Xymon avoids these performance bottlenecks by keeping most of the
>> ever-changing data in memory instead of on-disk, and by being implemented in
>> C rather than shell scripts.
>> 
>> 
>> I'd imagine if the above statements is correct the power of the server and
>> available memory is the only thing holding you back?
>> 
>> 
>> On Wed, Jul 8, 2009 at 2:31 PM, nico wrote:
>> 
>> 
>> Hi,
>> 
>> In my company, we are currently monitoring ~7000 devices (unix, microsoft,
>> network, firewall, storage) in 13 datacenters in different countries
>> 
>> (Spain, Italy, Germany, Belgium, France, Amsterdam, Belgium, UK).
>> 
>> The architecture is simple: one hobbit cluster using Heartbeat in each
>> datacenter (= 13 clusters) reporting to a central cluster by bbproxy and
>> 
>> some custom scripts to concatenate/consolidate the bb-hosts files
>> correctly.
>> 
>> My question is for the future, we need to support 20 000 devices and i
>> would like to know if hobbit is able to do that. (scalability).
>> 
>> Indeed, my company doesn¹t trust that it¹s possible, and they want to
>> move to HP BSM or EMC Smarts .... :-(
>> 
>> Any experience about that is welcome.
>> 
>> Regards,
>> Nico
>> 
>> To unsubscribe from the hobbit list, send an e-mail to
>> 
>> hobbit-unsubscribe (at) hswn.dk
>> 
>> 
>> 
> 
> _________________________________________________________________
> Hotmail® has ever-growing storage! Don¹t worry about storage limits.
> http://windowslive.com/Tutorial/Hotmail/Storage?ocid=TXT_TAGLM_WL_HM_Tutorial_
> Storage_062009
> To unsubscribe from the hobbit list, send an e-mail to
> hobbit-unsubscribe (at) hswn.dk
> 
>