You are viewing a plain text version of this content. The canonical link for it is here.
Posted to oak-dev@jackrabbit.apache.org by Angela Schreiber <an...@adobe.com> on 2012/03/22 10:39:14 UTC
Multiple JCR Repositories in a single MikroKernel instance (?)
hi
from some basic discussions at the last f2f in basel i got the
impression that we may have multiple independant JCR repositories
in the same MikroKernel instance.
if that was the case and not a misunderstanding from my side,
i would like us to discuss how those different JCR repositories
are being addressed.
yesterday i played around with moving the login to the SPI layer
(oak-core) and found out that i had the following parameters
at hand:
a) URL of the mikrokernel
b) Workspace node within that mikrokernel
but i didn't find out how i would have identified the target JCR
repository containing that workspace.
am i missing something or is this something we still have to
define?
kind regards
angela
Re: Multiple JCR Repositories in a single MikroKernel instance (?)
Posted by Michael Dürig <md...@apache.org>.
On 22.3.12 10:18, Jukka Zitting wrote:
> Hi,
>
> On Thu, Mar 22, 2012 at 11:08 AM, Michael Dürig<md...@apache.org> wrote:
>> I think this is not the case. If I'm not completely mistaken there will be
>> always an 1:1 correspondence between an MicroKernel instance and a
>> Repository instance.
>
> That's my thinking as well.
>
> In a cloud deployment the underlying cloud-scale storage backend can
> and probably will be used to host multiple repositories. Also in such
> cases I see us having multiple MicroKernels running on top of such a
> backend, one (cluster) for each logical repository.
>
> Thus there should be no need to distinguish between one repository and
> another in the oak-core level.
>
BTW the 1:1 correspondence was a major working assumption during the
last couple of months. This might be the reason it wasn't too explicitly
stated at the F2F.
Michael
Re: Multiple JCR Repositories in a single MikroKernel instance (?)
Posted by Angela Schreiber <an...@adobe.com>.
hi
it seems that tom got totally confused by my thread.
here some additional info:
what i meant was:
the mk-url was not only the identification of the storage
but would have (in case this is not yet the case) to identify
that part of the storage the contains the data of the
JCR repository that is being created.
when referring to the workspace name i was just omitting
some additional steps in my mind. what i meant was:
thus, identifying an item in the mk would in that case not only
consist of storage-address + wsp-Name + item-Path but
would rather have to be:
mk-url (being both storage and jcr-repo-identifier) + wsp-Name + item-Path
kind regards
angela
On 3/22/12 3:40 PM, Angela Schreiber wrote:
> hi
>
> On 3/22/12 11:18 AM, Jukka Zitting wrote:
>>> I think this is not the case. If I'm not completely mistaken there will be
>>> always an 1:1 correspondence between an MicroKernel instance and a
>>> Repository instance.
>>
>> That's my thinking as well.
>
> ok...
>
>> In a cloud deployment the underlying cloud-scale storage backend can
>> and probably will be used to host multiple repositories. Also in such
>> cases I see us having multiple MicroKernels running on top of such a
>> backend, one (cluster) for each logical repository.
>
> so, that would mean that it was just a matter of addressing. in
> other words: the URL use to create the MK instance must not only
> identify the store and the workspace but also the repository?
>
>> Thus there should be no need to distinguish between one repository and
>> another in the oak-core level.
>
> that was my assumption as well... i was just wondering about the
> mk implementation.
>
> kind regards
> angela
Re: Multiple JCR Repositories in a single MikroKernel instance (?)
Posted by Chetan Mehrotra <ch...@gmail.com>.
Hi Angela,
My understanding of the proposed design was that there would be two parts
1. MK Client - Each repository for a tenant would be using its repositories
MK Client stack to communicate with one server which provider the server
side implementation for MK layer
2. MK Server
Multiple Repository Instances through there Mk Client stacks would be
communicating with single MK Server. This server can be the store for all
the different repository instance (tenants). The MK Client would be using
some authentication keys based on which access to that section of store
would be allowed to it.
Chetan Mehrotra
On Thu, Mar 22, 2012 at 11:13 PM, Angela Schreiber <an...@adobe.com>wrote:
> hi tom
>
> i am not talking about multiple workspaces within a single
> jcr repository.
>
> i am talking about multiple jcr repositories within the same
> store each of which for example would have a 'default' workspace and
> a versionstore and so forth....
>
> kind regards
> angela
>
>
> On 3/22/12 5:19 PM, Thomas Mueller wrote:
>
>> Hi,
>>
>> in other words: the URL use to create the MK instance must not only
>>> identify the store and the workspace but also the repository?
>>>
>>
>> I thought we want to keep the data of all workspaces within the same
>> physical storage, so use the same MK instance for all workspaces within a
>> repository.
>>
>>
>> Regards,
>> Thomas
>>
>>
Re: Multiple JCR Repositories in a single MikroKernel instance (?)
Posted by Angela Schreiber <an...@adobe.com>.
hi tom
i am not talking about multiple workspaces within a single
jcr repository.
i am talking about multiple jcr repositories within the same
store each of which for example would have a 'default' workspace and
a versionstore and so forth....
kind regards
angela
On 3/22/12 5:19 PM, Thomas Mueller wrote:
> Hi,
>
>> in other words: the URL use to create the MK instance must not only
>> identify the store and the workspace but also the repository?
>
> I thought we want to keep the data of all workspaces within the same
> physical storage, so use the same MK instance for all workspaces within a
> repository.
>
>
> Regards,
> Thomas
>
Re: Multiple JCR Repositories in a single MikroKernel instance (?)
Posted by Thomas Mueller <mu...@adobe.com>.
Hi,
>in other words: the URL use to create the MK instance must not only
>identify the store and the workspace but also the repository?
I thought we want to keep the data of all workspaces within the same
physical storage, so use the same MK instance for all workspaces within a
repository.
Regards,
Thomas
Re: Multiple JCR Repositories in a single MikroKernel instance (?)
Posted by Angela Schreiber <an...@adobe.com>.
hi
On 3/22/12 11:18 AM, Jukka Zitting wrote:
>> I think this is not the case. If I'm not completely mistaken there will be
>> always an 1:1 correspondence between an MicroKernel instance and a
>> Repository instance.
>
> That's my thinking as well.
ok...
> In a cloud deployment the underlying cloud-scale storage backend can
> and probably will be used to host multiple repositories. Also in such
> cases I see us having multiple MicroKernels running on top of such a
> backend, one (cluster) for each logical repository.
so, that would mean that it was just a matter of addressing. in
other words: the URL use to create the MK instance must not only
identify the store and the workspace but also the repository?
> Thus there should be no need to distinguish between one repository and
> another in the oak-core level.
that was my assumption as well... i was just wondering about the
mk implementation.
kind regards
angela
Re: Multiple JCR Repositories in a single MikroKernel instance (?)
Posted by Jukka Zitting <ju...@gmail.com>.
Hi,
On Thu, Mar 22, 2012 at 11:08 AM, Michael Dürig <md...@apache.org> wrote:
> I think this is not the case. If I'm not completely mistaken there will be
> always an 1:1 correspondence between an MicroKernel instance and a
> Repository instance.
That's my thinking as well.
In a cloud deployment the underlying cloud-scale storage backend can
and probably will be used to host multiple repositories. Also in such
cases I see us having multiple MicroKernels running on top of such a
backend, one (cluster) for each logical repository.
Thus there should be no need to distinguish between one repository and
another in the oak-core level.
BR,
Jukka Zitting
Re: Multiple JCR Repositories in a single MikroKernel instance (?)
Posted by Michael Dürig <md...@apache.org>.
Hi,
On 22.3.12 9:39, Angela Schreiber wrote:
> hi
>
> from some basic discussions at the last f2f in basel i got the
> impression that we may have multiple independant JCR repositories
> in the same MikroKernel instance.
I think this is not the case. If I'm not completely mistaken there will
be always an 1:1 correspondence between an MicroKernel instance and a
Repository instance.
Michael
>
> if that was the case and not a misunderstanding from my side,
> i would like us to discuss how those different JCR repositories
> are being addressed.
>
> yesterday i played around with moving the login to the SPI layer
> (oak-core) and found out that i had the following parameters
> at hand:
>
> a) URL of the mikrokernel
> b) Workspace node within that mikrokernel
>
> but i didn't find out how i would have identified the target JCR
> repository containing that workspace.
>
> am i missing something or is this something we still have to
> define?
>
> kind regards
> angela