From: Joe Carlson <jwcarlson@lbl.gov>
Subject: Re: Row level security policy policy versus SQL
 constraints. Any performance difference?
To: Tom Lane <tgl@sss.pgh.pa.us>, Tomas Vondra <tomas.vondra@2ndquadrant.com>
References: <59E66BA8.6060009@lbl.gov>
 <993c32d2-1389-66aa-1368-ca44de370a3a@2ndquadrant.com>
 <14730.1508278004@sss.pgh.pa.us>
Cc: pgsql-performance@postgresql.org
Message-ID: <59E681C9.7050709@lbl.gov>
Date: Tue, 17 Oct 2017 15:18:49 -0700
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101
 Thunderbird/38.6.0
MIME-Version: 1.0
In-Reply-To: <14730.1508278004@sss.pgh.pa.us>
Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk
Sender: pgsql-performance-owner@postgresql.org

Thanks for your suggestions.

I had pretty much given up on this idea. At first, I had thought there 
would only be 2 or 3 different constraint cases to consider. I had 
thought of using distinct credentials for my connection and using RLS to 
give different cuts on the same table. The different policies could be 
established in advance and never touched.

But then it became clear that I actually would need a very large number 
of different restrictions on the tables - too many to create in advance. 
At this point it's easiest to apply constraints on each select rather 
than apply a policy every time.

Thanks,

Joe

On 10/17/2017 03:06 PM, Tom Lane wrote:
> Tomas Vondra <tomas.vondra@2ndquadrant.com> writes:
>> On 10/17/2017 10:44 PM, Joe Carlson wrote:
>>> What I was wondering is what is the performance differences between a
>>> row level security implementation:
>>> ...
>>> and an implementation where I add on the constraints as part of each
>>> select statement:
>> The main point of the RLS is enforcing an order in which the conditions
>> are evaluated.
> Yeah.  Because of that, I would *not* recommend RLS if you can equally
> well stick the equivalent conditions into your queries.  There is way
> too much risk of taking a serious performance hit due to a bad plan.
>
> An alternative you might consider, if simplifying the input queries
> is useful, is to put the fixed conditions into a view and query the
> view instead.  That way there's not an enforced evaluation order.
>
> 			regards, tom lane


-- 
Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance