| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Mailing Lists -> Oracle-L -> Re: High Availability - 99.999%
This is a multi-part message in MIME format.
------=_NextPart_000_0DF5_01C01404.9959FC90 Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
99.999% availability means 8.544 hours, say 8 1/2 hours of downtime a =
year! A machine with more than 1 processor, power-supply, Raid-x =
diskmirroring, seperate network connections etc. Could do the job.
Don't let any dba's, nor developers work on it - that's the most =
important thing -, and correct a problem immediately. So have a DBA =
stand by around the clock, for a down at 23:00, and up at 7:00 consumes =
all your 'accepted downtime'
Never upgrade the Oracle version, for this takes too long
etc. etc.
It should be possible to provide this availability without OPS. But if = it's really such a big issue, why not use it? We work with national = e-commerce, and most people sleep between say 3:00 and 6:00. Still we = use OPS. I can bring one server down to upgrade, correct etc. while the = other one still works. Switch the servers, and upgrade the second one. = Syncronize servers when all the time staying in the air. Machines are in = different locations, so trouble from the outside has no effect. (Some = years ago an airplane fell out of the sky near one of the locations...)
The only thing is, the sites are replicated immediate - it's not a hot = standby -. So if some ..... corrupts the database, the other one is = corrupted too! If this is possible, I think it's better to use a hot = standby with an update delay of say 1/2 an hour. You have to switch = manually, but recovery is not needed, so you can be back into business = quickly.
Eric Lansu=20
I am trying to understand the possible pieces going into providing =
99.999% availability in an Oracle/Sun environment. Everything I have =
read so far mentions using Oracle Parallel Server (OPS) to provide quick =
failover if a server (node) or instance fails. In your opinions are =
there other options besides OPS to provide this functionality?
=20
Thanks,
Nancy
=20
------=_NextPart_000_0DF5_01C01404.9959FC90 Content-Type: text/html;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content=3D"text/html; charset=3Diso-8859-1" =
http-equiv=3DContent-Type>
<META content=3D"MSHTML 5.00.3013.2600" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3DCourier size=3D2>99.999% availability means 8.544 =
hours, say 8 1/2=20
hours of downtime a year! A machine with more than 1 processor, =
power-supply,=20
Raid-x diskmirroring, seperate network connections etc. Could do the=20
job.</FONT></DIV>
<DIV><FONT face=3DCourier size=3D2>Don't let any dba's, nor developers =
work on it -=20
that's the most important thing -, and correct a problem immediately. So =
have a=20
DBA stand by around the clock, for a down at 23:00, and up at 7:00 =
consumes all=20
your 'accepted downtime'</FONT></DIV>
<DIV><FONT face=3DCourier size=3D2>Never upgrade the Oracle version, for =
this takes=20
too long</FONT></DIV>
<DIV><FONT face=3DCourier size=3D2>etc. etc.</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DCourier size=3D2>It should be possible to provide this =
availability without OPS. But if it's really such a big issue, why not =
use it?=20
We work with national e-commerce, and most people sleep between say =
3:00=20
and 6:00. Still we use OPS. I can bring one server down to upgrade, =
correct etc.=20
while the other one still works. Switch the servers, and upgrade =
the second=20
one. Syncronize servers when all the time staying in the =
air.=20
Machines are in different locations, so trouble from the outside has no =
effect.=20
(Some years ago an airplane fell out of the sky near one of the=20
locations...)</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DCourier size=3D2>The only thing is, the sites are =
replicated=20
immediate - it's not a hot standby -. So if some ..... corrupts the =
database,=20
the other one is corrupted too! If this is possible, I think it's better =
to use=20
a hot standby with an update delay of say 1/2 an hour. You have to =
switch=20
manually, but recovery is not needed, so you can be back into business=20
quickly.</FONT></DIV>
<DIV><FONT face=3DCourier size=3D2></FONT> </DIV>
<DIV><FONT face=3DCourier size=3D2>Eric Lansu</FONT> </DIV>
<BLOCKQUOTE=20
style=3D"BORDER-LEFT: #000000 2px solid; MARGIN-LEFT: 5px; MARGIN-RIGHT: =
0px; PADDING-LEFT: 5px; PADDING-RIGHT: 0px">
<DIV style=3D"FONT: 10pt arial">----- Original Message ----- </DIV>
<DIV=20
style=3D"BACKGROUND: #e4e4e4; FONT: 10pt arial; font-color: =
black"><B>From:</B>=20
<A href=3D"mailto:nancy.mccormick_at_sbti.com" =
title=3Dnancy.mccormick_at_sbti.com>Nancy=20
McCormick</A> </DIV>
<DIV style=3D"FONT: 10pt arial"><B>To:</B> <A =
href=3D"mailto:ORACLE-L_at_fatcity.com"=20
title=3DORACLE-L_at_fatcity.com>Multiple recipients of list ORACLE-L</A> =
</DIV>
<DIV style=3D"FONT: 10pt arial"><B>Sent:</B> Friday, 01 September 2000 =
00:18</DIV>
<DIV style=3D"FONT: 10pt arial"><B>Subject:</B> High Availability -=20
99.999%</DIV> <DIV><BR></DIV> <DIV><SPAN class=3D850445517-31082000><FONT face=3DArial size=3D2>I am =trying to=20
 |
 |