| Author |
Message
|
| bruce2359 |
 Posted: Mon Jul 26, 2010 8:31 am Post subject: |
 |
|
Poobah
Joined: 05 Jan 2008
Posts: 9482
Location: US: west coast, almost. Otherwise, enroute.
|
| Quote: |
| There is not whole lot of clarity in management. |
This is a perfect opportunity to start this conversation.
For our DR project, executives at my organization initially tasked IT to "...just back up everything; and figure out how to restore it." We had m/f's, Win, UNIX, 2500 desktops...
I asked them to identify the most-important application? They replied "...all of our applications are mission-critical." I asked them "Which of them is so important that it puts the very existence of the organization at risk?" (mumble, mumble, mumble, was their initial reply.)
"The revenue system." (They were finally getting some skin in the game.)
"What would be the next most-important?" Payroll came in #2, but only at a payroll cycle time. All other apps were (inevitably) assigned to #3 or below.
This conversation led to a more definitive DR plan - with sufficient focus to allow the IT folks to formulate a real business DR plan - one that the auditors bought into.
You have my sympathy. You and your DR plan are doomed to fail without some destination in mind (and on paper).
_________________
I like deadlines. I like to wave as they pass by.
ב''ה
Lex Orandi, Lex Credendi, Lex Vivendi. As we Worship, So we Believe, So we Live. |
|
| Back to top |
|
 |
| exerk |
| Posted: Mon Jul 26, 2010 8:34 am Post subject: |
|
|
Jedi Council
Joined: 02 Nov 2006
Posts: 6339
|
| shashivarungupta wrote: |
| ...So, Nodes can be geographically spread over the network... |
I suspect geographical in the sense jeevan is using the term may be as spread as Michigan is to Bangalore, hence the query as to whether HACMP will work in that case.
_________________
It's puzzling, I don't think I've ever seen anything quite like this before...and it's hard to soar like an eagle when you're surrounded by turkeys. |
|
| Back to top |
|
|
| Vitor |
| Posted: Mon Jul 26, 2010 8:37 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| bruce2359 wrote: |
| Quote: |
| There is not whole lot of clarity in management. |
This is a perfect opportunity to start this conversation. |
| bruce2359 wrote: |
| I asked them to identify the most-important application? They replied "...all of our applications are mission-critical." |
Now I've heard that a few times. My, slightly less professional, retort to a similar management response was "all of them? Including the one that sends out the discount vouchers every sale time?".
It could be claimed to be a mission-critical part of the sales process, but I just wanted to hear the justification. Which was not forthcoming.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| shashivarungupta |
| Posted: Mon Jul 26, 2010 8:52 am Post subject: |
|
|
Grand Master
Joined: 24 Feb 2009
Posts: 1343
Location: Floating in space on a round rock.
|
| exerk wrote: |
| shashivarungupta wrote: |
| ...So, Nodes can be geographically spread over the network... |
I suspect geographical in the sense jeevan is using the term may be as spread as Michigan is to Bangalore, hence the query as to whether HACMP will work in that case. |
Why do you suspect ?  Reason(s) ?
_________________
*Life will beat you down, you need to decide to fight back or leave it. |
|
| Back to top |
|
|
| shashivarungupta |
| Posted: Mon Jul 26, 2010 8:56 am Post subject: |
|
|
Grand Master
Joined: 24 Feb 2009
Posts: 1343
Location: Floating in space on a round rock.
|
| exerk wrote: |
| shashivarungupta wrote: |
| ...So, Nodes can be geographically spread over the network... |
I suspect geographical in the sense jeevan is using the term may be as spread as Michigan is to Bangalore, hence the query as to whether HACMP will work in that case. |
By the way: I did edit my post in 1st page regarding dependence on n/w topology before you posted.  
_________________
*Life will beat you down, you need to decide to fight back or leave it. |
|
| Back to top |
|
|
| exerk |
| Posted: Mon Jul 26, 2010 8:59 am Post subject: |
|
|
Jedi Council
Joined: 02 Nov 2006
Posts: 6339
|
| shashivarungupta wrote: |
| exerk wrote: |
| shashivarungupta wrote: |
| ...So, Nodes can be geographically spread over the network... |
I suspect geographical in the sense jeevan is using the term may be as spread as Michigan is to Bangalore, hence the query as to whether HACMP will work in that case. |
Why do you suspect ? Reason(s) ? |
Because jeevan wrote: "...We can not use the MI features as GPFS does not work in geographically spread area..."
_________________
It's puzzling, I don't think I've ever seen anything quite like this before...and it's hard to soar like an eagle when you're surrounded by turkeys. |
|
| Back to top |
|
|
| jeevan |
| Posted: Mon Jul 26, 2010 10:53 am Post subject: |
|
|
Grand Master
Joined: 12 Nov 2005
Posts: 1432
|
| bruce2359 wrote: |
| Quote: |
| There is not whole lot of clarity in management. |
This is a perfect opportunity to start this conversation.
For our DR project, executives at my organization initially tasked IT to "...just back up everything; and figure out how to restore it." We had m/f's, Win, UNIX, 2500 desktops...
I asked them to identify the most-important application? They replied "...all of our applications are mission-critical." I asked them "Which of them is so important that it puts the very existence of the organization at risk?" (mumble, mumble, mumble, was their initial reply.)
"The revenue system." (They were finally getting some skin in the game.)
"What would be the next most-important?" Payroll came in #2, but only at a payroll cycle time. All other apps were (inevitably) assigned to #3 or below.
This conversation led to a more definitive DR plan - with sufficient focus to allow the IT folks to formulate a real business DR plan - one that the auditors bought into.
You have my sympathy. You and your DR plan are doomed to fail without some destination in mind (and on paper). |
This is true in general but I am in the process of creating DR plan and while I am in the process, I never hope for failure. If I do not believe, I do not work on the project.
Evey DR project should be owned by business. Or even we can generalise any IT or non It project not owned by business will fail. In that context DR plan /project is not exception. But personally, in my case, I make it clear to my manager that this is business take and this is IT ( my take). Thus I am hoping, we will come out with a good and workable plan.
My aim in this post is to explore the technological dimension of the IT DR plan, feasibility advantage/disadvantages etc. I am not a business manager and do not have much control over it though I can definitely try to influence wchich I am doing already.
Last edited by jeevan on Mon Jul 26, 2010 12:56 pm; edited 1 time in total |
|
| Back to top |
|
|
| mqjeff |
| Posted: Mon Jul 26, 2010 10:58 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
Neither HACMP nor MI are strictly suited for long distance, DR based recovery.
A BACKUP queue manager, available in v6 and later, is. |
|
| Back to top |
|
|
| jeevan |
| Posted: Mon Jul 26, 2010 11:42 am Post subject: |
|
|
Grand Master
Joined: 12 Nov 2005
Posts: 1432
|
| mqjeff wrote: |
Neither HACMP nor MI are strictly suited for long distance, DR based recovery.
A BACKUP queue manager, available in v6 and later, is. |
I just discovered that SRDF is another technologies that could be feasible for long distance DR.
However, i do not know it much and need to explore further. |
|
| Back to top |
|
|
| zpat |
| Posted: Mon Jul 26, 2010 11:50 am Post subject: |
|
|
Jedi Council
Joined: 19 May 2001
Posts: 5867
Location: UK
|
The main problem with MQ multi-instance as opposed to HA/CMP is the lack of a same IP address takeover.
We can't change all our client apps to have multiple IP addresses for connectivity and so restoring the queue manager on a different IP address is of limited value.
Hopefully IBM will provide a means to involve a script at takeover time that could be used to drive a DNS update or something like that. But that is still a slow process.
Currently we are stuck with having HA/CMP clusters at both primary and DR sites (we have hot DR). I would love to have a single HA over two sites but don't know a way to make it work.
Even using data replication like SRDF won't move the original IP address. So the delay on failover would be significant as well - manually updating DNS and waiting for propagation of the change.
Last edited by zpat on Mon Jul 26, 2010 11:52 am; edited 2 times in total |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Mon Jul 26, 2010 11:51 am Post subject: |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9482
Location: US: west coast, almost. Otherwise, enroute.
|
You should look into GDPS, too. It's more mature and robust technology.
_________________
I like deadlines. I like to wave as they pass by.
ב''ה
Lex Orandi, Lex Credendi, Lex Vivendi. As we Worship, So we Believe, So we Live. |
|
| Back to top |
|
|
| Vitor |
| Posted: Mon Jul 26, 2010 12:00 pm Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| zpat wrote: |
| So the delay on failover would be significant as well - manually updating DNS and waiting for propagation of the change. |
This is an important general point. Key differences between HA & DR:
HA tends to be a single application stack, DR tends to be most or all of an estate
HA required recovery times tend to be measured in seconds, DR measured in hours
HA tends to be heavily automated (due to the short recovery times), DR is more manual.
So the choice of technology is influenced by the recovery times required.
I'm also still waiting for an answer to my question "what counts as a distaster if the loss of a datacentre isn't?" 
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Mon Jul 26, 2010 12:04 pm Post subject: |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9482
Location: US: west coast, almost. Otherwise, enroute.
|
| Quote: |
| I'm also still waiting for an answer to my question "what counts as a distaster if the loss of a datacentre isn't?" |
As we move slowly up the cost side of a cost-benefit analysis, a single data center image is sometimes referred to a "single point of failure."
_________________
I like deadlines. I like to wave as they pass by.
ב''ה
Lex Orandi, Lex Credendi, Lex Vivendi. As we Worship, So we Believe, So we Live. |
|
| Back to top |
|
|
| jeevan |
| Posted: Mon Jul 26, 2010 12:48 pm Post subject: |
|
|
Grand Master
Joined: 12 Nov 2005
Posts: 1432
|
| bruce2359 wrote: |
| Quote: |
| I'm also still waiting for an answer to my question "what counts as a distaster if the loss of a datacentre isn't?" |
As we move slowly up the cost side of a cost-benefit analysis, a single data center image is sometimes referred to a "single point of failure." |
Guys, I am still working on it and talking to people. unitl I talk to my manager and ask him hard quesitons like what is the management vision of the DR, I will not have a clear answer. Also, I am talking to differtent folks Unix /midrange who are involved in managing Servers and storage(SAN or other storage) so things are evolving as I move.
The cost benefits analysis is important but it comes little later once we know the business goal and choose the technology ( as business goal influnce the technologies to be deployed), then comes cost component. |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Mon Jul 26, 2010 1:07 pm Post subject: |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9482
Location: US: west coast, almost. Otherwise, enroute.
|
What!? Analyze FIRST, then design? This will never catch on. 
_________________
I like deadlines. I like to wave as they pass by.
ב''ה
Lex Orandi, Lex Credendi, Lex Vivendi. As we Worship, So we Believe, So we Live. |
|
| Back to top |
|
|
|
|
