Using JCR to implement an -indexed and searchable- event log

[Guillaume Belrose <ka...@gmail.com>]

'Hi all,

I've been implementing a light weight media asset management service
using JackRabbit which will soon be rolled out in production to
customers. The service provides a RESTful HTTP interface to allow
clients to organize media assets in "folders" which are implemented as
JCR nodes with custom properties. I would like to be able to keep
track of important events such as creation and deletion of assets so
that those actions can be indexed and searched for by an
administrator.

I was thinking of storing each core event as a jcr node with
properties to record time-stamp and various type information and I am
wondering if this is suitable approach.I am aware that JackRabbit
struggles to scale if a given node contains a large (i.e. over 10000 )
number of children nodes. I guess to go around this problem, I could
store events by using a node structure modelled after the date (year
--> month --> day) when events occur. This would certainly limit the
number of nodes to be stored at any given level.

I am wondering if people in the community have used JackRabbit to
implement such a searchable event log feature before.

Thanks in advance,

Guillaume.

Re: Using JCR to implement an -indexed and searchable- event log

[Guillaume Belrose <ka...@gmail.com>]

Ha ha,
Glad to hear this.
Merci,
Guillaume

On Tue, Oct 26, 2010 at 4:39 PM, Bertrand Delacretaz
<bd...@apache.org> wrote:
> On Tue, Oct 26, 2010 at 5:29 PM, Guillaume Belrose <ka...@gmail.com> wrote:
>> ...I was thinking of storing each core event as a jcr node with
>> properties to record time-stamp and various type information and I am
>> wondering if this is suitable approach.I am aware that JackRabbit
>> struggles to scale if a given node contains a large (i.e. over 10000 )
>> number of children nodes. I guess to go around this problem, I could
>> store events by using a node structure modelled after the date (year
>> --> month --> day) when events occur. This would certainly limit the
>> number of nodes to be stored at any given level....
>
> Way to go, this allows you to have an arbitrary number of event nodes,
> and makes browsing/purging events very easy using a JCR explorer.
>
> We're using similar structures in several places in Day's cq5 product.
>
> -Bertrand
>

Re: Using JCR to implement an -indexed and searchable- event log

[Bertrand Delacretaz <bd...@apache.org>]

On Tue, Oct 26, 2010 at 5:29 PM, Guillaume Belrose <ka...@gmail.com> wrote:
> ...I was thinking of storing each core event as a jcr node with
> properties to record time-stamp and various type information and I am
> wondering if this is suitable approach.I am aware that JackRabbit
> struggles to scale if a given node contains a large (i.e. over 10000 )
> number of children nodes. I guess to go around this problem, I could
> store events by using a node structure modelled after the date (year
> --> month --> day) when events occur. This would certainly limit the
> number of nodes to be stored at any given level....

Way to go, this allows you to have an arbitrary number of event nodes,
and makes browsing/purging events very easy using a JCR explorer.

We're using similar structures in several places in Day's cq5 product.

-Bertrand