| |
|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
  |
|
| LogWriteIntegrity.... should I pick SingleWrite or Triple? |
« View previous topic :: View next topic » |
| Author |
Message
|
| PeterPotkay |
 Posted: Tue Jul 14, 2015 4:26 pm Post subject: LogWriteIntegrity.... should I pick SingleWrite or Triple? |
 |
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
|
| Back to top |
|
 |
| zpat |
| Posted: Tue Jul 14, 2015 9:58 pm Post subject: |
|
|
Jedi Council
Joined: 19 May 2001
Posts: 5866
Location: UK
|
I've been using single write for years without any issues. On a modern SAN the writes are guareented to happen. In fact we synchronously write over two sites with SRDF. There are significant performance gains from using it.
_________________
Well, I don't think there is any question about it. It can only be attributable to human error. This sort of thing has cropped up before, and it has always been due to human error. |
|
| Back to top |
|
|
| Andyh |
| Posted: Wed Jul 15, 2015 12:13 am Post subject: |
|
|
Master
Joined: 29 Jul 2010
Posts: 239
|
As the referenced article indicates, there would typically only be significant performance gains from SingleWrite in an MQ environment with little concurrency. With SRDF then you're pretty much guaranteed relatively high I/O latency (can't do much about the speed of light), and hence achieving high throughput in that sort of environment typically requires a fair degree of concurrency.
Take a trivial application doing a repeating sequence of MQPUT ; MQCMIT ; MQGET ; MQCMIT. Each iteration of the app will require 2 log forces (one for each commit) and thus will take a minimum of 2 I/O latencies (=>SingleWrite), but would typically take 4 I/O latencies with TripleWrite.
A number of customers have demonstrated this kind of benchmark and have concluded there are significant gains from SingleWrite, however unless your production workload follows a similar pattern then it's not a very fair comparison.
If you were to run concurrent 50 instances of the same application it would be a better representation of a more typical production workload, and in that environment I would expect minimal differences between SingleWrite and TripleWrite.
As the referenced article suggests, you have to really know what you're doing to know whether SingleWrite is really safe. It's related with the atomicity of writing 4KB aligned blocks to disk (regardless of the severity of any failure) and so you'd have to fully understand the entire I/O stack from top to bottom in some detail to safely deploy SingleWrite. In the vast majority of cases TripleWrite would be the better option, for example do you really trust that this option would be revisited in the underlying file-system or SAN was changed. |
|
| Back to top |
|
|
| zpat |
| Posted: Wed Jul 15, 2015 7:01 am Post subject: |
|
|
Jedi Council
Joined: 19 May 2001
Posts: 5866
Location: UK
|
You would be amazed at the weaknesses that exist in application design from the "new generation" of developers who have little knowledge of transactional integrity.
Compared to that, I don't lose sleep over Single Write on our SAN !!
_________________
Well, I don't think there is any question about it. It can only be attributable to human error. This sort of thing has cropped up before, and it has always been due to human error. |
|
| Back to top |
|
|
|
|
|
| |
|
Page 1 of 1 |
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
|
|
