another reason for hierarchical design?

Dear list

I have been told many times that a well designed directory is as 
flat as possible. Hierarchical directory often bring a lot of trouble 
moving entries around. The reason for hierarchical design exist mainly 
for replication and ACL control reason. While the later reason doesn't 
really exist with OpenLDAP thanks to its extremely flexible ACL, the
directory server of my choice, the only remaining reason for
hierarchical design seems to be replication.

Now I am facing a choice to choose a flat structure or a hierarchical. I
have drafted two of my choices and attached to this email.

Every reference I read seems to suggest me use the flat structure but I
cannot solve this problem: we have two special case:

1) combined search for person by the company they are in is one of the
most frequent search users do, e.g.

	(sn=Wang,o=Seven Sea) -> find all Mr Wang working for Seven Sea
	(l=ShangHai,o=ABB)   -> find all ABB's Shang Hai contact persons

2) the compan name is updated very frequently, for various reasons. One
reason is spelling mistakes (the people who filles in Company Name first
time are often not native speakers of English, due to our real situation
and we administrators cannot change this.) There are other practical
reasons too.

3) often, each company nave many contact persons in our directory.

With our situation it seems there is a problem we cannot solve with flat
structure: If a company name is being updated, we must locate all
contact persons in this company by doing a search, using search filter
like (o=ABB), and update every one of them.

This problem is avoided with hierarchical structure. See the picture I
attached. If the company name change, the person node under this company
do not have to update at all. The only difference is we write filters
like this:
	(sn=Wang,o:dn:=Seven Sea)
	(l=ShangHai,o:dn:=ABB)

Each time the company name is changed, we do a modrdn operation to the
company node, while do not need to update the contact persons beneath
it.

While my practical case suggested maybe hierarchical structure is
better, I was warned a lot of times hierarchical structure rarely fits
better, especially my reason of using hierarchical structure is not the
"standard" reason given by experts (i.e. replication). So I think I
should consult experts on the list rather than head on some stupid
incorrect idea of mine.

I can think of one downside of using hierarchical structure in my case,
that is if someone move from company A to company B, a modrdn operation
is needed rather than a simple LDAP modify operation that replace the
value of "o". I hope I didn't miss other issues.

Thank you in advance for your time and professional suggestions!

Best regards

Dear list

I have been told many times that a well designed directory is as 
flat as possible. Hierarchical directory often bring a lot of trouble 
moving entries around. The reason for hierarchical design exist mainly 
for replication and ACL control reason. While the later reason doesn't 
really exist with OpenLDAP thanks to its extremely flexible ACL, the
directory server of my choice, the only remaining reason for
hierarchical design seems to be replication.

Now I am facing a choice to choose a flat structure or a hierarchical. I
have drafted two of my choices and attached to this email.

Every reference I read seems to suggest me use the flat structure but I
cannot solve this problem: we have two special case:

1) combined search for person by the company they are in is one of the
most frequent search users do, e.g.

	(sn=Wang,o=Seven Sea) -> find all Mr Wang working for Seven Sea
	(l=ShangHai,o=ABB)   -> find all ABB's Shang Hai contact persons

2) the compan name is updated very frequently, for various reasons. One
reason is spelling mistakes (the people who filles in Company Name first
time are often not native speakers of English, due to our real situation
and we administrators cannot change this.) There are other practical
reasons too.

3) often, each company nave many contact persons in our directory.

With our situation it seems there is a problem we cannot solve with flat
structure: If a company name is being updated, we must locate all
contact persons in this company by doing a search, using search filter
like (o=ABB), and update every one of them.

This problem is avoided with hierarchical structure. See the picture I
attached. If the company name change, the person node under this company
do not have to update at all. The only difference is we write filters
like this:
	(sn=Wang,o:dn:=Seven Sea)
	(l=ShangHai,o:dn:=ABB)

Each time the company name is changed, we do a modrdn operation to the
company node, while do not need to update the contact persons beneath
it.

While my practical case suggested maybe hierarchical structure is
better, I was warned a lot of times hierarchical structure rarely fits
better, especially my reason of using hierarchical structure is not the
"standard" reason given by experts (i.e. replication). So I think I
should consult experts on the list rather than head on some stupid
incorrect idea of mine.

I can think of one downside of using hierarchical structure in my case,
that is if someone move from company A to company B, a modrdn operation
is needed rather than a simple LDAP modify operation that replace the
value of "o". I hope I didn't miss other issues.

Thank you in advance for your time and professional suggestions!

Best regards