Transparent data encryption support as an extension

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Transparent data encryption support as an extension

Haribabu Kommi-2
Hi Hackers,

I read many mail discussions in supporting data at rest encryption support in
PostgreSQL.

I checked the discussions around full instance encryption or tablespace or
table level encryption. In my observation, all the proposals are trying to modify
the core code to support encryption.

I am thinking of an approach of providing tablespace level encryption support
including WAL using an extension instead of changing the core code by adding
hooks in xlogwrite and xlogread flows, reorderbuffer flows and also by adding
smgr plugin routines to support encryption and decryption of other pages. 

Definitely this approach does't work for full instance encryption.

Any opinions/comments/problems in evaluating the encryption with an extesnion
approach?

Regards,
Haribabu Kommi
Fujitsu Australia
Reply | Threaded
Open this post in threaded view
|

Re: Transparent data encryption support as an extension

Masahiko Sawada
On Fri, Apr 12, 2019 at 6:34 PM Haribabu Kommi <[hidden email]> wrote:

>
> Hi Hackers,
>
> I read many mail discussions in supporting data at rest encryption support in
> PostgreSQL.
>
> I checked the discussions around full instance encryption or tablespace or
> table level encryption. In my observation, all the proposals are trying to modify
> the core code to support encryption.
>
> I am thinking of an approach of providing tablespace level encryption support
> including WAL using an extension instead of changing the core code by adding
> hooks in xlogwrite and xlogread flows, reorderbuffer flows and also by adding
> smgr plugin routines to support encryption and decryption of other pages.
>
> Definitely this approach does't work for full instance encryption.
>
> Any opinions/comments/problems in evaluating the encryption with an extesnion
> approach?
>

The discussion[1] of similar proposal might be worth to read. The
proposal was adding hook in BufferSync, although for differential
backup purpose.

[1] https://www.postgresql.org/message-id/20051502087457@...

Regards,

--
Masahiko Sawada
NIPPON TELEGRAPH AND TELEPHONE CORPORATION
NTT Open Source Software Center