Re: [postgis-devel] JDBC & WKB - 400% overhead - Mailing list pgsql-jdbc

From Markus Schaber
Subject Re: [postgis-devel] JDBC & WKB - 400% overhead
Date
Msg-id 421C99E1.4070107@logi-track.com
Whole thread Raw
Responses Re: [postgis-devel] JDBC & WKB - 400% overhead
List pgsql-jdbc
[X-Post to better place for discussion]

Hi, David,

dblasby@openplans.org schrieb:
> I was checking to see if the WKB (postgis 1.0) was actually transiting
> "in binary" from the server (postgresql 8) to a JDBC (also postgresql
> 8) client.
>
> ..
> ResultSet rs2 = st.executeQuery("select 5::float8,asBinary('POINT(0 0)')");
> rs2.next();
> byte[] bs = rs2.getBytes(2);
> assertTrue(bs.length == 21);
>
> The WKB representation of a 'POINT(0 0)' is 21 bytes long (1 byte for
> xdr/ndr flag, 4 bytes for type, and 2 8-byte doubles = 21 bytes).
>
> If you step through the executeQuery() command in the postgresql driver,
> you'll eventually end up in PGStream#ReceiveTupleV3().  This is where
> the actual data is received from the database.
>
> It clearly receives 84 bytes of data from the server.  This is the bytea
> text representation:
> \001\001\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000
>
> The ResultSet#getBytes() function converts this to a byte[].

Maybe your postgresql driver only uses text transfer up to now.
asBinary() returns a byte array, so this should be transferred more
efficiently.

Can you try the following query:

select 5::float8,asBinary('POINT(0 0)')::bytea

Maybe the current 8.0-310 build has fixed this. But the pgsql-jdbc list
could be the better place to ask this question, IMHO, so I forward it to
there.


Markus

--
markus schaber | dipl. informatiker
logi-track ag | rennweg 14-16 | ch 8001 zürich
phone +41-43-888 62 52 | fax +41-43-888 62 53
mailto:schabios@logi-track.com | www.logi-track.com


Attachment

pgsql-jdbc by date:

Previous
From: Ken Geis
Date:
Subject: Re: Performance tweaks
Next
From: Sergei Georgiev
Date:
Subject: Socket timeouts (probably)