| Author |
Message
|
| Mitra |
 Posted: Fri Jun 03, 2011 7:29 am Post subject: Suggest some High Availabity Methods |
 |
|
Voyager
Joined: 21 Mar 2011
Posts: 78
|
Hi , Can any one suggest some HA methods for Failover As our company doesn't want to have VCS as it is expensive
2. Not ready to have Heart beat Clustering as we have no Support.
3. Not agreeing with MQ V7 Multi Instance Qmgrs as Clients also has to Change to V7.
Plz suggest some good High availabilty Solutions for Failover
Thanks |
|
| Back to top |
|
 |
| Vitor |
| Posted: Fri Jun 03, 2011 7:46 am Post subject: Re: Suggest some High Availabity Methods |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| Mitra wrote: |
| Plz suggest some good High availabilty Solutions for Failover |
So your company wants a robust, reliable, cost-free HA solution? Do post a link when you find one as I think all our clients would be interested...
There are any number of posts in this forum from people who have tried to use a WMQ Cluster as an HA solution and have discovered that it has a number of limitations, many of which rule it out as an HA solution (becuase that's not what a WMQ cluster is intended to be).
But if your company is prepared to live with this cheap and inadequate solution you could try that. Make sure you have hardcopy of your statement detailing the problems signed by all the stakeholders kept safe somewhere. Because when you hit an actual problem & try to use it, someone's going to whine at you that it's not all that good & some data's gone AWOL & you'll need something to defend yourself with. You'll be able to assemble this document by distilling all the posts from people along the lines of "we're using a WMQ cluster for HA and <problem> has happened. How do we resolve?" and "In this HA scenario, how would a WMQ cluster cope?". The answers to both classes of post are "you can't" and "it doesn't", in that order.
HA for clients? A CCDT and proper code in the applications to deal with dropped connections, including reconnection.
Obviously this is only HA for the WMQ part of your stack. You'll need to kludge something else for the databases, app servers and so forth.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| mqjeff |
| Posted: Fri Jun 03, 2011 7:50 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
Get a clear and complete statement of all of the failure conditions that your HA solution is supposed to relieve.
That is, is your HA solution supposed to allow for Clients to reconnect to a failed qmgr quickly?
Is your HA solution supposed to allow for the recovery of 'in progress' messages?
These kinds of things. Get a clear and complete list of REQUIREMENTS.
Then assemble a list of the COST of each of those failed situations. How much money will the company lose if one of those situations occurs for five minutes? ten minutes? ten hours? ten days?
These potential costs then equal the VALUE you are receiving from your current HA solution.
You can then use this VALUE to either justify the COST of your current HA solution, or to determine how much you can afford on your new HA solution.
Don't forget to include the cost of migrating to a new HA solution when figuring out what you can afford. |
|
| Back to top |
|
|
| Mitra |
| Posted: Fri Jun 03, 2011 7:55 am Post subject: Re: Suggest some High Availabity Methods |
|
|
Voyager
Joined: 21 Mar 2011
Posts: 78
|
| Vitor wrote: |
| Mitra wrote: |
| Plz suggest some good High availabilty Solutions for Failover |
So your company wants a robust, reliable, cost-free HA solution? Do post a link when you find one as I think all our clients would be interested...
There are any number of posts in this forum from people who have tried to use a WMQ Cluster as an HA solution and have discovered that it has a number of limitations, many of which rule it out as an HA solution (becuase that's not what a WMQ cluster is intended to be).
But if your company is prepared to live with this cheap and inadequate solution you could try that. Make sure you have hardcopy of your statement detailing the problems signed by all the stakeholders kept safe somewhere. Because when you hit an actual problem & try to use it, someone's going to whine at you that it's not all that good & some data's gone AWOL & you'll need something to defend yourself with. You'll be able to assemble this document by distilling all the posts from people along the lines of "we're using a WMQ cluster for HA and <problem> has happened. How do we resolve?" and "In this HA scenario, how would a WMQ cluster cope?". The answers to both classes of post are "you can't" and "it doesn't", in that order.
HA for clients? A CCDT and proper code in the applications to deal with dropped connections, including reconnection.
Obviously this is only HA for the WMQ part of your stack. You'll need to kludge something else for the databases, app servers and so forth. |
HI Vitor,
Thanks for the Reply ,
I already suggested the WMQ Clustering and explained them that it's not gonna provide HA for us(Just the Load balancing even it is not achieved as we have just 1 qmgr to be clustered for right now). So it was ruled out.
Then i gave the solution of MI in V7
but they are saying that the Clients have to be in V7 and they are sticking to the limitations of MI
(Limitations of multi-instance queue managers include:
Automatic client reconnect is not supported by WebSphere MQ classes for Java (even if client changes to V7 as we are using Java programs)
•A standby instance must be restarted manually when a failover occurs)
so they are not ready to restart the Stand by instance manually when ever we have failover.
So are there any HA solutions other than these. |
|
| Back to top |
|
|
| Mitra |
| Posted: Fri Jun 03, 2011 8:01 am Post subject: |
|
|
Voyager
Joined: 21 Mar 2011
Posts: 78
|
| mqjeff wrote: |
Get a clear and complete statement of all of the failure conditions that your HA solution is supposed to relieve.
That is, is your HA solution supposed to allow for Clients to reconnect to a failed qmgr quickly?
Is your HA solution supposed to allow for the recovery of 'in progress' messages?
These kinds of things. Get a clear and complete list of REQUIREMENTS.
Then assemble a list of the COST of each of those failed situations. How much money will the company lose if one of those situations occurs for five minutes? ten minutes? ten hours? ten days?
These potential costs then equal the VALUE you are receiving from your current HA solution.
You can then use this VALUE to either justify the COST of your current HA solution, or to determine how much you can afford on your new HA solution.
Don't forget to include the cost of migrating to a new HA solution when figuring out what you can afford. |
Thanks mqjeff ,
The thing they want is they want to get rid of VCS which we are using for HA/Failover now, as it is Costly .
They want the solution with out changing anything. |
|
| Back to top |
|
|
| mqjeff |
| Posted: Fri Jun 03, 2011 8:06 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
| Mitra wrote: |
The thing they want is they want to get rid of VCS which we are using for HA/Failover now, as it is Costly .
They want the solution with out changing anything. |
It is costly to spend $100 on a new shirt. but if I am at an airport and my luggage has been lost and I have to be at a client in an hour and I have spilled coffee all over the shirt I'm wearing... then the VALUE of that $100 is WORTH the cost.
To say something is expensive means that the value it is providing is greater than the cost.
You can't say that until you can DEMONSTRATE the value.
If you just want "free HA", then try Mr. Google. But it won't help until you can get the client to demonstrate the value they receive.
And you can't "get rid of VCS" "without changing anything". You have to CHANGE VCS to GET RID of it. |
|
| Back to top |
|
|
| Vitor |
| Posted: Fri Jun 03, 2011 8:09 am Post subject: Re: Suggest some High Availabity Methods |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| Mitra wrote: |
| So are there any HA solutions other than these. |
Yes - VCS, HeartBeat, and any number of others. Most of which cost money.
As my most worthy associate correctly points out, the cost of any HA solution is mitigated by the cost to the business of downtime. The cheapest solution is to sit and wait for the original system to be repaired. Clearly if it's a hardware fault that can take some time. Your management will (unless they're totally crazy) say "we can't afford to be down that long" at which point you have a baseline amount!.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| Vitor |
| Posted: Fri Jun 03, 2011 8:11 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| Mitra wrote: |
| They want the solution with out changing anything. |
They want a robust, cost-free, fully automated HA solution that doesn't require any changes to implement?
I apologise for my previous post - your management are totally crazy.
Perhaps they'd like the moon on a string as well? Perhaps a herd of unicorns they can breed for the horns? 
_________________
Honesty is the best policy.
Insanity is the best defence.
Last edited by Vitor on Fri Jun 03, 2011 8:13 am; edited 1 time in total |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Fri Jun 03, 2011 8:12 am Post subject: |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9469
Location: US: west coast, almost. Otherwise, enroute.
|
To-date, "free HA" has proven to be neither free nor HA.
_________________
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: Fri Jun 03, 2011 8:13 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| bruce2359 wrote: |
| To-date, "free HA" has proven to be neither free nor HA. |
In line with the well-accepted principle of "you get what you pay for"
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| Mitra |
| Posted: Fri Jun 03, 2011 8:41 am Post subject: |
|
|
Voyager
Joined: 21 Mar 2011
Posts: 78
|
| Vitor wrote: |
| bruce2359 wrote: |
| To-date, "free HA" has proven to be neither free nor HA. |
In line with the well-accepted principle of "you get what you pay for" |
I have doubt, like as it is stated in the Limitations of MI do we need to restart the Standby Instance every time the Fail over occurs in order to make the stand by an Active Instance. If so how can we say that the Stand by will take over the Active Instance work when Active Instance fails ? Plz clarify my doubt.
Thanks, |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Fri Jun 03, 2011 8:44 am Post subject: |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9469
Location: US: west coast, almost. Otherwise, enroute.
|
I'm not sure what are you expecting? You can script the startup of the next standby. You can script the startup of the next standby after that.
_________________
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 |
|
|
| Mitra |
| Posted: Fri Jun 03, 2011 8:48 am Post subject: |
|
|
Voyager
Joined: 21 Mar 2011
Posts: 78
|
| bruce2359 wrote: |
| I'm not sure what are you expecting? You can script the startup of the next standby. You can script the startup of the next standby after that. |
ok thank you and here is my another Question as we are using Java programs can't MI Support for this at all.
Source: Automatic client reconnect is not supported by MQ classes for Java
What is meant by MQ classes of Java here?
and what is this
•For a non-JMS client, set the mqclient.ini environment variable DefRecon to set reconnection options. Use a CCDT to connect to a queue manager. If the client is to connect to a multi-instance queue manager, provide the network addresses of the active and standby queue manager instances in the CCDT.
is Java different from JMS
plzz help me as i not having Any knowledge in Java at all |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Fri Jun 03, 2011 9:00 am Post subject: |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9469
Location: US: west coast, almost. Otherwise, enroute.
|
| Mitra wrote: |
| bruce2359 wrote: |
| I'm not sure what are you expecting? You can script the startup of the next standby. You can script the startup of the next standby after that. |
ok thank you and here is my another Question as we are using Java programs can't MI Support for this at all.
Source: Automatic client reconnect is not supported by MQ classes for Java
What is meant by MQ classes of Java here? |
Did you search (here or google) for automatic client reconnection?
MQ classes for Java is a language implementation - like COBOL or C or JMS.
_________________
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 |
|
|
| Mitra |
| Posted: Fri Jun 03, 2011 9:09 am Post subject: |
|
|
Voyager
Joined: 21 Mar 2011
Posts: 78
|
| Mitra wrote: |
So as we are using Java Programs we can't use MI  |
No, no, no.
The reference you cited (and Table 1 Supported clients) clearly indicates that program access to connection options is not supported using WebSphere MQ (base) classes for Java.
It does NOT state that you can't use MI if you use WebSphere MQ classes for Java. |
|
| Back to top |
|
|
|
|
