From: Luke H. Crouch
Date: August 10 2004 7:07pm
Subject: RE: gigabit performance...?
List-Archive: http://lists.mysql.com/cluster/341
Message-Id: <075C170BC7219F439DD3FE8C880648C1AB0A90@red-ivy.red-man.com>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

okay...have a little more info about our gigabit performance.

set some of the linux parameters for gigabit networks according to this =
page:
http://datatag.web.cern.ch/datatag/howto/tcp.html

when testing copying a huge file from one machine in the cluster to =
another...
-rw-r--r--    1 root     root     481302559 Aug 10 08:12 gmargin1.sql =
(481 MB)
Tue Aug 10 08:12:10 CDT 2004
Tue Aug 10 08:12:25 CDT 2004	(15 seconds to copy file)

481MB/15s =3D 32MB/s or 256 Mbit.


looks okay...higher than 100Mbit at least. now onto the DB speed...

mysql> describe gmargin;
+-----------------+---------------+------+-----+---------+-------+
| Field           | Type          | Null | Key | Default | Extra |
+-----------------+---------------+------+-----+---------+-------+
| gm_quote        | char(4)       |      | MUL |         |       |
| gm_cust_no      | char(8)       |      | MUL |         |       |
| gm_part         | char(7)       | YES  |     | NULL    |       |
| gm_family       | char(3)       |      | MUL |         |       |
| gm_sub_family   | char(3)       |      |     |         |       |
| gm_ticket       | char(8)       | YES  |     | NULL    |       |
| gm_qty          | decimal(12,4) |      |     | 0.0000  |       |
| gm_list_price   | decimal(12,4) |      |     | 0.0000  |       |
| gm_unit_net     | decimal(12,4) |      |     | 0.0000  |       |
| gm_unit_cost    | decimal(12,4) |      |     | 0.0000  |       |
| gm_disc_cat     | char(1)       | YES  |     | NULL    |       |
| gm_price_source | char(1)       | YES  |     | NULL    |       |
| gm_pd           | int(11)       |      | MUL | 0       |       |
| gm_user         | char(3)       | YES  |     | NULL    |       |
+-----------------+---------------+------+-----+---------+-------+
14 rows in set (0.01 sec)

I calculate the row size as 99 bytes.

mysql> select count(*) from gmargin;
+----------+
| count(*) |
+----------+
|  4355766 |
+----------+
1 row in set (26.64 sec)

and, as I understand, this sent all 4,355,766 rows over the network to =
the mysql server, so 400,355,766 bytes, right? does this seem like a =
reasonable time? I'm not sure how much time it is taking the mysql =
server to count up all the rows...?

400MB/26.6s =3D 15MB/s or 120 Mbit.

basically, we're interested in all the settings we can change to max out =
the performance on our cluster. is there any kind of optimization guide =
for the cluster?

thanks again,
-L