Unsolved
This post is more than 5 years old
341 Posts
0
3639
HP-UX11iV3 native multipathing vs PowerPath
Hi,
Just wondering if anyone has any working experience of HP-UX11iV3's native-multipathing feature which, according to the HP documentation provides both path-failover and load balancing.
- Does this mean that PowerPath is now redundant on this platform?
- Are there any advantages or valid reasons for installing PP on this platform?
- Can PP co-exist with the native multipathing, or must it be disabled... how?
any info appreciated
CB
Just wondering if anyone has any working experience of HP-UX11iV3's native-multipathing feature which, according to the HP documentation provides both path-failover and load balancing.
- Does this mean that PowerPath is now redundant on this platform?
- Are there any advantages or valid reasons for installing PP on this platform?
- Can PP co-exist with the native multipathing, or must it be disabled... how?
any info appreciated
CB
bodnarg
385 Posts
0
July 27th, 2007 06:00
The best answer would probably come from someone within EMC that could maybe make a technical argument for using PowerPath in addition or in place of the multi-pathing supplied by the OS.
Navin-Swn_9
66 Posts
1
July 27th, 2007 09:00
"The mass storage stack in HP-UX 11i v3 uses that agility to provide transparent multi-pathing. In other words, if a LUN has multiple lunpaths, I/O requests can be transparently distributed across all available lunpaths to the LUN, using a choice of load balancing algorithms. This eliminates the need for add-on multi-pathing products."
i can email you the pdf. test mail at - navinu@gmail.com
Conor
341 Posts
0
August 9th, 2007 00:00
PowerPath 5.1.0 for HP-UX supports HP-UX 11i v3.0 only with minimum qualified level September 2007 (0709) HP-UX patch.
available on http://powerlink.emc.com
bodnarg
385 Posts
0
August 9th, 2007 04:00
Mabro1
666 Posts
0
August 9th, 2007 05:00
I noticed that your question as yet has not been answered. If your issue is urgent and you need Technical assistance then I would advise you to contact EMC technical support or your local EMC customer support representative.
Should you need further assistance on this please contact me.
Kind regards,
Mark
Message was edited by:
Mark Browne
Conor
341 Posts
0
August 9th, 2007 07:00
That is the main features that make PowerPath a better option than the native failover...
dynamox
2 Intern
2 Intern
•
20.4K Posts
0
August 9th, 2007 12:00
xe2sdc
2 Intern
2 Intern
•
2.8K Posts
0
August 10th, 2007 02:00
Now they both are using the WWID (something like a WWN but bound to a specific device and not to the port) to identify multiple paths to a given disk and handle multipathing.
Is PowerPath better ?? I can't run tests on the subject so I can't answer the question. I can say that for an HP-UX administrator the "NGMS" will be a revolution (or maybe an evolution) of the existing commands while PP is something "outside" the OS.
Anesh-nr-uW
10 Posts
0
January 28th, 2009 16:00
what I find is native load balancing gives (round-robin bydefault )..this is what we are using. but I am not sure which policy is better. With power path we always used adaptive load balancing policy.
two questions:
which one is better ? (Power path or native load balancing)
if native load balancing which policy is better for DMX devices ?
regards
Anesh
MarkF4
39 Posts
1
January 29th, 2009 16:00
What type of storage system are you connecting to? If you were using previously using Adaptive with PowerPath, that would suggest that you weren't connecting to a Symmetrix/DMX but instead a third-party array.
However, when it comes to load balancing policy in a DMX environment "Symmetrix optimization" should give better performance than "Round Robin".
To quote the PowerPath 5.1 product guide, (Symmetrix optimization ensures) "I/O requests are routed to paths based on an algorithm that takes into account path load and logical device priority. Load is a function of the number, size, priority, and type of I/O queued on each path". Contrast to that Round Robin, in which "I/O requests are assigned to each available path in rotation."
So if the host has 2 paths, 1 busy path and 1 quiet path, "Symmetrix Optimization" will drive any new I/O's to the quiet path whereas "Round Robin" will continue spreading I/O evenly between the 2 paths.
If native load balancing only allows "Round Robin", I'd suggest that PowerPath would utilize the paths better and therefore be the "better load balancing software" even without taking it's other features into account.
Mark
Anesh-nr-uW
10 Posts
0
June 29th, 2011 21:00
and we CX, DMX, hitachi
Anesh-nr-uW
10 Posts
0
June 29th, 2011 21:00
thanks mark..
this i still a debate now we moved Redhat with native where it provides just round robin.
since last two years we are using native..as such we don't see performance issues..
but I am unable to prove that powerpath is better
Brion2
154 Posts
0
June 30th, 2011 05:00
Hi Anesh. It really depends on how you are testing load balancing. If you are using a single host with dual HBAs running an I/O load using PP and then using round robin to 2 or more storage ports without any other SAN activity or environment variability, then I would expect to see similar performance between SymmOPT and CLAROpt versus round robin. If you would to have a discussion on test scenarios that will show the PP advantage, I'd be happy to discuss.
Besides load balancing, we have a number of other benefits. A few examples are proactive failover on VNX/CX NDU (non-disruptive upgrade), load-based path testing, management (new PowerPath Viewer), advanced features (non-disruptive LUN migration and encryption).
Thanks, Brion
Brion Stultz
Corporate System Engineer - PowerPath/UIM
EMC2 Corporation
SKT2
2 Intern
2 Intern
•
1.3K Posts
0
August 10th, 2011 07:00
Brion,
what you meant "load-based path testing"?
Brion2
154 Posts
0
August 10th, 2011 08:00
Hi. PowerPath tests dead paths, active paths, and idle paths. PowerPath doesn't perform tests on a fixed schedule under all I/O conditions. That is, when there is heavy I/O on the system, PowerPath doesn't need to perform a path test. When data is returned on a read request or an acknowledgement is returned on a write request, we know that the path is good. So, there is no need to burden the host or network with additional path tests. Does that help?
-Brion