FYI, for database discussion. Ignore other stuff.
-----Original Message-----
From: aklug@aklug.org [mailto:aklug@aklug.org]=20
Sent: Thursday, March 23, 2006 10:25 AM
To: aklug digest users
Subject: aklug Digest V5 #56
aklug Digest Wed, 22 Mar 2006 Volume: 05 Issue: 056
In This Issue:
Re: Question regarding databases
RE: OFF-TOPIC: bridging a T1 to ethernet=20
Re: Question regarding databases
Re: Question regarding databases
re: removing large file
Re: removing large file
Interesting app for top & ps users
Re: Question regarding databases
Ping times
----------------------------------------------------------------------
From: "Neil Moomey" <neil@neilmoomey.com>
Subject: Re: Question regarding databases
Date: Wed, 22 Mar 2006 09:25:13 -0900
PostgreSQL is better but MySQL is good enough and is more popular with
ISPs.=20
You can use PhpMyAdmin to load and edit your data. SQL Server is too=20
expensive for the web and doesn't run on Linux. Access is limited to 16
concurrent users. Oracle is overkill. I recommend MySQL for your
purpose.
Neil=20
------------------------------
Subject: RE: OFF-TOPIC: bridging a T1 to ethernet=20
Date: Wed, 22 Mar 2006 09:25:21 -0900
From: "Leif Sawyer" <lsawyer@gci.com>
Gah, stupid OWA top-posting..
=20
Anyway, you've got a couple of options:
=20
One, you could go with a Cisco router and an outboard pix. This is the
'normal' method.
=20
Two, you could go with a PC-based router with a T1 card. That'll give
you T1-Ethernet plus
an iptables-based firewall. XORP is a good place to start.
=20
Three, you could go with a non-pix based commercial firewall. Juniper,
Resilience, and a handfull
of others make a firewall with built-in T1 termination.
________________________________
From: aklug-bounce@aklug.org on behalf of Thomison, Lee
Sent: Wed 2006-03-22 9:23 AM
To: aklug@aklug.org
Subject: OFF-TOPIC: bridging a T1 to ethernet=20
Hey folks,
It's a bit off topic, but still techie....
We have a T1 link to one of our partners. Right now the T1 goes into a
CSU/DSU, and comes out a V.35 into a cisco router that routes some of
the subnets and locations in the company
We now want to put a pix between us and the partner. Pix's don't have
V.35 modules. Nor do they have anything that'll accept a T1. Pix's
want Ethernet. Period.
This has to be a common requirement, but so far the only solution I've
found is routers (e.g. cisco, adtran 3205 , kentrox Q2200) with T1's on
the outside and ethernet on the inside. Seems overkill to me, not to
mention complicating the communications path and giving Murphy an order
of magnitude or two more to screw things over.
What's normally done in this circumstance?
Thanks,
---------
To unsubscribe, send email to <aklug-request@aklug.org>
with 'unsubscribe' in the message body.
------------------------------
From: Geoffrey Wright <geoff.wright@pangomedia.com>
Subject: Re: Question regarding databases
Date: Wed, 22 Mar 2006 10:15:57 -0900
I'd second Matthew's thoughts about this subject.
We do a lot of develop against quite a few databases in my office. =
=3D20
In terms of OSS offerings, we use both PostgreSQL and MySQL.
The former is more mature and more featureful. This latter is (at =3D20
least historically) easier to get started with.
The usual complaints levied against PostgreSQL relative to MySQL are:
1 - It doesn't run well on Windows.
2 - It doesn't have as nice a collection of client interfaces.
3 - It last the raw select speed of MySQL.
The first complaint was addressed quite some time ago. PostgreSQL =3D20
now runs as happily on Windows as it does on *NIX, and actually runs =
=3D20
quite a bit faster on OS X than MySQL does.
MySQL does have a richer collection of dev tools than PostgreSQL =3D20
does, but there are still several good ones for PostgreSQL. Stuff we
=3D20=3D
use around here includes: phpPgAdmin, Navicat (also nice for MySQL) =
=3D20
pgAdmin III. I really like the Navicat tool -- it has native clients
=3D20=3D
for both Mac OS and Windows, but I don't believe they have one for =3D20
Linux yet.
MySQL is definitely scorchingly fast in some situations -- especially
=3D20=3D
where you are doing simple selects using the MyISAM table type. But =
=3D20
this only really matters if you're doing huge stuff. And you pay a =
=3D20
stiff price in losing RI. Still though, MySQL is great in certain =3D20
situations where speed on simple selects is key.
I guess the summary is that MySQL and PostgreSQL are both decent =3D20
choices. PostgreSQL _might_ be a little harder to get started with, =
=3D20
but it's probably worth the hassle for the features and maturity you =
=3D20
gain. That said we're developed some pretty nice apps against MySQL,
=3D20=3D
so you probably won't go wrong there, either.
If course the fact that Oracle keeps buying key components of MySQL =
=3D20
does make me a little bit nervous... :)
//glw
On Mar 21, 2006, at 9:28 PM, Matthew Schumacher wrote:
> Tony wrote:
>> Hi Everyone:
>>
>> I have a question. I am setting up a database and I
>> want this database
>> to be accessible via a web browser to both input the
>> data and retrieve
>> the data.
>>
>> My choices are SQL, MySQL, Oracle, microsoft Access,
>> and who knows what
>> else.
>>
>> Does any one have a recommendation as to which
>> database software is best
>> for a database warehouse?
>>
>
> Here are my thoughts on those databases:
>
> 1. MsSQL, not a bad database really, but it's expensive, a pain to
> backup, needs it's on box to run on, and has a few security =3D20
> issues. On
> the flip side, it's one of the easiest to manage.
>
> 2. MySQL. It does store and fetch your data, but almost every real
> feature in it is an after thought and where you do have support some
> features they are almost always limited to the common use of that
> feature. See my examples below.
>
> 3. Oracle. Larry Ellison is the greediest man on the planet.
>
> 4. Microsoft Access is not a real database. Run away, run away!!
>
> I recommend postgres. I have been using it for a few years and it =
=3D20
> has a
> bunch more features the mysql, and in many ways compares with =3D20
> oracle and
> MSSQL server.
>
> Here are two quick examples of where a mysql feature is an =3D20
> afterthought
> and postgres gets it right:
>
>
> Example 1
>
=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D=
3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D
=3D
=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D=
3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D
=3D3D=3D
=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D=
3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D20
> =3D3D=3D3D
> Mysql does not allow you to use now() as the default value of a =3D20
> column.
> =3D46rom their docs:
>
> "The DEFAULT clause specifies a default value for a column. With one
> exception, the default value must be a constant; it cannot be a =3D20
> function
> or an expression. This means, for example, that you cannot set the
> default for a date column to be the value of a function such as =
NOW=3D20
> () or
> CURRENT_DATE. The exception is that you can specify =3D20
> CURRENT_TIMESTAMP as
> the default for a TIMESTAMP column. See Section 11.3.1.1, =
=3D93TIMESTAMP
> Properties as of MySQL 4.1=3D94. "
>
> So they work around this major shortcoming by giving people the
> CURRENT_TIMESTAMP keyword for the timestamp column so at least the
> people wanting an automatic timestamps are happy. Basically they add
> just enough support to do the most common thing.
>
> Postgres, allows all of the functions in a create table statement.
=3D20=3D
> This
> is legit in postgres:
>
> create table test_tab (data varchar(20), timestamp timestamp default
> now() - interval '32.56 minutes');
>
>
> Example 2
>
=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D=
3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D
=3D
=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D=
3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D
=3D3D=3D
=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D=
3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D3D=3D20
> =3D3D=3D3D
> Mysql allows you use the "INSERT ... ON DUPLICATE KEY UPDATE" which
=3D20=3D
> gets
> around the problem of inserting or updating depending on whether the
> data is there, where postgres gives you a complete rule system which
> allows you to:
>
> CREATE RULE update_rule AS ON UPDATE TO test_tab WHERE EXISTS (select
> key from test_tab) DO INSTEAD insert....
>
> It may seem that postgres is more complex, and in some ways it is, =
=3D20
> but I
> would much rather have a complete rule system then support for a few
> common things everyone wants to do.
>
> Hopefully this information will help,
>
> schu
>
>
>
> ---------
> To unsubscribe, send email to <aklug-request@aklug.org>
> with 'unsubscribe' in the message body.
>
>
------------------------------
Date: Wed, 22 Mar 2006 15:07:19 -0800 (PST)
From: Tony <vze2jy85@yahoo.com>
Subject: Re: Question regarding databases
Hi Neil:
Thanks for your reply.
Tony
Neil Moomey wrote:
> PostgreSQL is better but MySQL is good enough and is
more popular with=20
> ISPs. You can use PhpMyAdmin to load and edit your
data. SQL Server=20
> is too expensive for the web and doesn't run on
Linux. Access is=20
> limited to 16 concurrent users. Oracle is overkill.
I recommend MySQL=20
> for your purpose.
>
> Neil
>
>
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around=20
http://mail.yahoo.com=20
------------------------------
Date: Wed, 22 Mar 2006 17:35:14 -0800 (PST)
From: Kurt Brendgard <brendgard@yahoo.com>
Subject: re: removing large file
It does sound as if something has gone wrong.=20
It almost sounds as if the system needs more ram to
keep track of what sectors need to be dealt with,
maybe try add a stick long enough to delete it? Or
maybe if you have a real memory hog of an app running,
try turning ot off while deleting. Linux is better
than Windows with using smaller amounts of ram, but
there is only so much that can be expected.
If the box has enough ram, it's also possible that the
memory and/or mb are getting flakey, trying to access
ram in a spot where it's starting to flake out. Might
try swapping out with some new ram stick to see if
that works.
If you can't get it deleted any other way, there is a
last resort trick to try. Try copying a much smaller
file over the top of it, or opening it up to write,
overwriting whats there, and write very little. makes
the memory requirments smaller in deleting it. Last
resort though, if something goes wrong, it could make
it much worse.
Kurt
Date: Wed, 22 Mar 2006 03:10:45 +0000
From: Douglas McIntosh=20
Subject: removing large file
Somehow I imported a 1.6 Gb rpm file and now I need to
delete it.
Rm and mv, which work on smaller files, both segfault
with the
big file. I use the Reiser file system and all
utilities are up to
date. Kernel 2.6.12.
I have enough space remaining on my drive so this is
not a critical
problem.
I stupidly thought that the linux maximum file size of
2 Gb meant
that slightly smaller files were ok, but maybe not.
Comments, anyone?
Douglas
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around=20
http://mail.yahoo.com=20
------------------------------
Date: Wed, 22 Mar 2006 18:36:38 -0800
From: Mac Mason <mac@cs.hmc.edu>
Subject: Re: removing large file
On Wed, Mar 22, 2006 at 05:35:14PM -0800, Kurt Brendgard wrote:
> It almost sounds as if the system needs more ram to
> keep track of what sectors need to be dealt with,
> maybe try add a stick long enough to delete it? Or
> maybe if you have a real memory hog of an app running,
> try turning ot off while deleting. Linux is better
> than Windows with using smaller amounts of ram, but
> there is only so much that can be expected.
I'm no filesystems guru, but I really doubt it takes much ram at all
to delete a file. It's just a tree traversal, right? Even with a
collossal filesystem (=3D> large tree elements), that shouldn't be too
hard. Shouldn't be any more memory-intensive than writing the file to
disk the first time, and that worked, right?
In the worst case, you'll just hit swap space; if you're truly out of
ram, that's a kernel panic, not a segfault (unless rm was written by
morons, but it wasn't).=20
If you feel like it, it could be fun to figure out where rm is
segfaulting; attach a debugger and go to town.
> If the box has enough ram, it's also possible that the
> memory and/or mb are getting flakey, trying to access
> ram in a spot where it's starting to flake out. Might
> try swapping out with some new ram stick to see if
> that works.
Better yet, run memtest86. My favorite way to do this is to download a
Knoppix CD and use that. It'll also provide an easy way to fsck your
filesystems. Bad ram could easily cause your problem.
=20
> If you can't get it deleted any other way, there is a
> last resort trick to try. Try copying a much smaller
> file over the top of it, or opening it up to write,
> overwriting whats there, and write very little. makes
> the memory requirments smaller in deleting it. Last
> resort though, if something goes wrong, it could make
> it much worse.
Or, mount the filesystem from Knoppix and delete the file from there.
Or (last last ditch) back up everything else, burn it down, and
restore from backups. But there's no way this is necessary.
--Mac
--=20
Julian "Mac" Mason mac@cs.hmc.edu
Computer Science '06 (310)-882-8068
Harvey Mudd College
-- Attached file included as plaintext by Ecartis --
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (FreeBSD)
iD8DBQFEIgmz1AphoTGXiN0RAiPpAJ48mycOSZ17fBwxYQWj7xyRterHwgCdGHh2
oB8xfmZtWvBopAqJ6zVlWic=3D
=3DQ3Y4
-----END PGP SIGNATURE-----
------------------------------
From: Greg Madden <pabi@gci.net>
Subject: Interesting app for top & ps users
Date: Wed, 22 Mar 2006 19:02:10 -0900
If you like gui versions of amin tools, Qps combines the info you get
from=20
using top & ps. It can scroll through the full list of processes,=20
something I had not figured out how to do with top, see the full output.
Debian has a package 'apt-get install qps'. It is a qt app.
http://qps.kldp.net/
--=20
Greg Madden
------------------------------
Subject: Re: Question regarding databases
Date: Thu, 23 Mar 2006 08:23:29 -0900
From: "Thomison, Lee" <ThomisonL@ci.anchorage.ak.us>
Well, first a disclaimer: I am not and do not want to be a database
guru. My first instinct when confronted with something that requires
database work is to point to someone, anyone, else and then/either/or
run and hide.
That being said, database work is like housework. Like it or not, if
your gonna be in this biz you're gonna have to get dirty sometimes.
Here's my take:
MS Access is a database only in the sense that it's a place to put and
organize data. It has a screen and reporting functionality that muggles
seem to find comforting even though I generally found it frustrating.
It's ok for quick-and-dirty-and-temporary, but as always, nothing is
more permanent than temporary. Calling it a 'single-user' database may
be most accurate. It does not serve data out, someone always has to be
logged in and running Access for anyone to get the data. Very 80's.
MS SQL Server is a Microsoft product, so my inclination is "how good can
it be?". Nonetheless, many people who would not otherwise be
ms-partisan have good things to say about it. I suspect that it's like
most commercial software tho. It's more about what you can't do than
what you can, and how much more money they can get you to cough up to
increase the 'what you can do' side of the fence.
Oracle used to be the system of choice, but they've gotten really
(shamelessly, even) greedy, and their licensing has gotten arcane and
convoluted. Frankly, I can't imagine any practical scenario that would
make it worth the hassle anymore.
Postgres has a small but devoted following. With the usual disclaimer
that I've never actually used it, it does look like it was done with a
craftsman's approach, that is, do it right the first time, make sure it
works and works well.
Mysql is very popular. Probably 80% of the OSS that can work with
database backends it's with mysql. It's what my company has (and hence
I have) decided to use just for that reason, it's what our software is
designed to work with. I have found it non-intuitive to set up and use;
YMMV.
Good luck. Let us know what you decide and why.
------------------------------
From: "W.D.McKinney" <deem@wdm.com>
Date: Thu, 23 Mar 2006 01:38:49 +0000
Subject: Ping times
Wonder when ACS & GCI will ever peer?
-Dee
------------------------------
End of aklug Digest V5 #56
**************************
---------
To unsubscribe, send email to <aklug-request@aklug.org>
with 'unsubscribe' in the message body.
---------
To unsubscribe, send email to <aklug-request@aklug.org>
with 'unsubscribe' in the message body.
Received on Thu Mar 23 10:59:18 2006
This archive was generated by hypermail 2.1.8 : Thu Mar 23 2006 - 10:59:18 AKST