| Author |
Message
|
| George Carey |
 Posted: Fri Sep 12, 2008 12:24 pm Post subject: Circular vs Linear logging debate |
 |
|
Knight
Joined: 29 Jan 2007
Posts: 500
Location: DC
|
Does this best practices paragraph from IBM dated July 16 2008 finally put a nail in the coffin of linear vs circular logging debate ???
http://www.ibm.com/developerworks/websphere/library/techarticles/0807_hsieh/0807_hsieh.html?S_TACT=105AGX54&S_CMP=B0724&ca=dnw-929
The top 15 WebSphere MQ best practices
...
Use the appropriate logging mechanism
A common misconception in the MQ community is that linear logging is more appropriate for production-level systems. In addition to the circular logging benefits, linear logging only incrementally provides object-recovery abilities, and it comes with the cost of regular maintenance to prevent file systems from filling. If the MS03 SupportPac is set up to run regularly, a copy of objects can be replayed from those scripts, minimizing the need for linear logging. Circular-logging queue managers require less maintenance and have less risk of crashing due to using up disk space.
GTC
_________________
"Truth is ... grasping the virtually unconditioned",
Bernard F. Lonergan S.J.
(from book titled "Insight" subtitled "A Study of Human Understanding") |
|
| Back to top |
|
 |
| jeevan |
| Posted: Fri Sep 12, 2008 1:55 pm Post subject: Re: Circular vs Linear logging debate |
|
|
Grand Master
Joined: 12 Nov 2005
Posts: 1432
|
| George Carey wrote: |
Does this best practices paragraph from IBM dated July 16 2008 finally put a nail in the coffin of linear vs circular logging debate ???
http://www.ibm.com/developerworks/websphere/library/techarticles/0807_hsieh/0807_hsieh.html?S_TACT=105AGX54&S_CMP=B0724&ca=dnw-929
The top 15 WebSphere MQ best practices
...
Use the appropriate logging mechanism
A common misconception in the MQ community is that linear logging is more appropriate for production-level systems. In addition to the circular logging benefits, linear logging only incrementally provides object-recovery abilities, and it comes with the cost of regular maintenance to prevent file systems from filling. If the MS03 SupportPac is set up to run regularly, a copy of objects can be replayed from those scripts, minimizing the need for linear logging. Circular-logging queue managers require less maintenance and have less risk of crashing due to using up disk space.
GTC |
Although, principally, Linear provides the ability to recover the
"damaged" objects, but the question is how often do we have this situation? Also, with the availability of saveqmgr, we can keep a copy of latest definition, and thus recreating a fresh new mq objects and setup does not take more than a few minitues should such incident occurs.
Tthough we can set up a mechanism to make linear logging cleanup automatic, still it does not add any value. However, there is obvious performance impact.The support pack : MP7E talks about this.
MP7E: WebSphere MQ for Windows V5.3 -
Circular and Linear Logging
However, I would love to hear from Gurus
Last edited by jeevan on Fri Sep 12, 2008 2:17 pm; edited 1 time in total |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Fri Sep 12, 2008 1:57 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
| george carey wrote: |
| it comes with the cost of regular maintenance to prevent file systems from filling. |
As we all know this too can be automated. (crontab etc...)
As your production boxes are monitored I would have expected that you got alerts if despite your automated jobs your file system threatened to blow up....
I am sorry to say but for some of us media recovery is a must, and adding a little disk space cheaper than having to enable producer app logging so as to be able to resend messages  
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| George Carey |
| Posted: Fri Sep 12, 2008 2:10 pm Post subject: contact admin raid |
|
|
Knight
Joined: 29 Jan 2007
Posts: 500
Location: DC
|
If a little extra disk is not a problem I would guess the price for a SAN or contact admin raid shelf or two from the likes of Network Appliance corp would be a must hardware solution to media issues and forget about your automation scripts. Of course that would move the admin skill set from MQ to NetApp filer admins.
_________________
"Truth is ... grasping the virtually unconditioned",
Bernard F. Lonergan S.J.
(from book titled "Insight" subtitled "A Study of Human Understanding") |
|
| Back to top |
|
|
| bower5932 |
| Posted: Fri Sep 12, 2008 2:17 pm Post subject: |
|
|
Jedi Knight
Joined: 27 Aug 2001
Posts: 3023
Location: Dallas, TX, USA
|
From the listserver....I wasn't sure whether to put the poster's name:
| Quote: |
Geeze, I hope not! This article has it's good points but is only a
summary. For example, this particular paragraph fails to mention the
cases where linear logging is absolutely required. As such it only hints
at the much greater complexity underlying this issue. The only "debate"
about linear vs. circular is what the default should be in the absence of
any defining business requirements. However business requirements for
speed or reliability usually dictate what the implementation should be.
Unfortunately, this article does not provide any insight as to how one
might make that decision or what the cost/benefit analysis would include.
Someone without knowledge of all the factors involved must rely on the
author's opinion of what is generally "best" and go with the
one-size-fits-all solution.
The article also promotes at least one anti-pattern in the naming of
cluster queues as TO.<QMGR> which works well only where clusters do not
overlap. Even if you do not consider this an anti-pattern, it is
certainly not the "best" practice (despite it being used in the IBM
manuals to illustrate cluster construction).
In the one-step-forward-two-steps-back category, the article says to
*consider* locking down administrative access on *production* systems.
Please do not interpret this as a statement that IBM's official position
is that locking down admin access is optional on any system. I'm happy
the authors thought to include something on security but appalled at the
message that was sent. Again, this section only hints at the underlying
complexity and gets some of the information wrong. If someone followed
the security "Best Practices" from this article to the letter, their
systems would still be wide open to anonymous administrative access yet
there is no hint of this and no link to find more information.
The best practice that is missing is that no checklist can adequately
substitute for deep skills and the ability to synthesize new solutions to
unique problems. To some extent "Best Practice" documents tend to inhibit
the development of such deep skills. Some are better than others but this
effect is especially true with documents that do not hint at underlying
complexity, guide the reader to additional reading or provide decision
criteria between alternatives.
My article "Embracing cultural change in the WebSphere MQ community" (
http://is.gd/2hOY ) was written in part as a response to the Top 15
article. Over and over again I go to assignments where a customer has
followed somebody's "Best" practice document to the letter and found
themselves in deep trouble. Once these things get started, it is
especially hard to change them. If someone comes up with a better "best"
practice, it has to compete against all the instances of the older "best"
practice out there. Many times the least-best "best" practice wins out
simply due to inertia. Especially if the established "best" practice is
easy to implement and grossly oversimplifies a complex issue such as
choosing the right logging configuration or securing the messaging
network. |
|
|
| Back to top |
|
|
| fjb_saper |
| Posted: Fri Sep 12, 2008 2:19 pm Post subject: Re: contact admin raid |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
| George Carey wrote: |
| If a little extra disk is not a problem I would guess the price for a SAN or contact admin raid shelf or two from the likes of Network Appliance corp would be a must hardware solution to media issues and forget about your automation scripts. Of course that would move the admin skill set from MQ to NetApp filer admins. |
And guess who is on the spot and responsible for fixing the blow up? The MQAdmin! The NetApp filer admins will argue => Not my problem!! 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| Michael Dag |
| Posted: Fri Sep 12, 2008 3:05 pm Post subject: |
|
|
Jedi Knight
Joined: 13 Jun 2002
Posts: 2607
Location: The Netherlands (Amsterdam)
|
my 2 cents...
Queues still get damaged and recovery of a damaged Queue is no fun... and isn't about the definitions (aka MS03 backup), it's about the data that was often still in it! 
_________________
Michael
MQSystems Facebook page |
|
| Back to top |
|
|
| jeevan |
| Posted: Fri Sep 12, 2008 3:07 pm Post subject: Re: contact admin raid |
|
|
Grand Master
Joined: 12 Nov 2005
Posts: 1432
|
| fjb_saper wrote: |
| George Carey wrote: |
| If a little extra disk is not a problem I would guess the price for a SAN or contact admin raid shelf or two from the likes of Network Appliance corp would be a must hardware solution to media issues and forget about your automation scripts. Of course that would move the admin skill set from MQ to NetApp filer admins. |
And guess who is on the spot and responsible for fixing the blow up? The MQAdmin! The NetApp filer admins will argue => Not my problem!! |
Is it just a matter of a few more disk space or performance also comes into picture? who is responsible is one thing but what is good is another question. Question what is better from the point of business not from how easy is it for MQ admin. |
|
| Back to top |
|
|
| George Carey |
| Posted: Fri Sep 12, 2008 3:15 pm Post subject: Scope of Field |
|
|
Knight
Joined: 29 Jan 2007
Posts: 500
Location: DC
|
...
| Quote: |
My article "Embracing cultural change in the WebSphere MQ community" (
http://is.gd/2hOY )
|
Excellent article T-rob ... not sure if you have a degree in I/T or Sociology(perhaps both). I would only add to his concerns over cultural obdurance and resistance to change of the 'commonly held view' in this case for 'Best Practices for MQ', would be the additional concern for cultural myopism ... or not looking outside the (MQ only) box for new solutions/paradigms.
http://www.ibm.com/developerworks/websphere/techjournal/0809_mismes/0809_mismes.html
_________________
"Truth is ... grasping the virtually unconditioned",
Bernard F. Lonergan S.J.
(from book titled "Insight" subtitled "A Study of Human Understanding") |
|
| Back to top |
|
|
| George Carey |
| Posted: Fri Sep 12, 2008 3:24 pm Post subject: Scope of Field |
|
|
Knight
Joined: 29 Jan 2007
Posts: 500
Location: DC
|
... dup error ... oops what logging should I have been using ??l 
_________________
"Truth is ... grasping the virtually unconditioned",
Bernard F. Lonergan S.J.
(from book titled "Insight" subtitled "A Study of Human Understanding") |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Fri Sep 12, 2008 4:03 pm Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
http://www.ibm.com/developerworks/websphere/library/techarticles/0712_dunn/0712_dunn.html
Here's another nail:
| Quote: |
| If you need to be able to forward recover queue data following a failure or recover from media failure of the device containing the log you must use linear logging if you are dependent on WebSphere MQ to provide this level of protection for you. An alternative strategy is to use disk mirroring to mirror the log device. This is often a facility provided by a SAN. In this case you could use circular logging. |
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
|
|
