hmm....i did not know about the issue with ATA and code 19, i have two CX600 with ATA and FC running on that code ...would you please point me to some white papers on this, cause right now i have my ATA luns spread all over the place.

Thanks

As far as best practices for assigning SPs, I would say plan on load balancing them. Don't worry about type of storage managed, just try to balance the actual I/O load as best you can.

One thing to watch out for is with the ATA drives though. You have to assign all the LUNs on a single ATA RAID group to a single SP. If you don't it will cause some serious trespassing (performance problems). I'm not sure if they fixed things in FLARE 24 to force you to do this, but in 19 you can do it wrong and pay the price if you aren't paying attention.

If you want something beyond what NaviSphere Analyzer gives you, there are a couple of potions, but the only one I'm familiar with is Performance manager in ECC. This will let you analyze the life out of every aspect of your storage array for as far back as you want to keep the data.

The other thing you can do to get a better picture than just the last 24 hours is use NaviCLI to pull the NAR file from an array on a 24 hour schedule. You can then look at individual NAR files, or use the interface to merge multiple files into a longer window.

Hope this helps some.

Hi Guys,

The reason all LUN's in an ATA RAID group should be assigned to a single SP is that the ATA drives themselves are not dual-ported like the FC drives. This means that the drive cannot service requests from both SP's simultaneously like the FC drives can.

If you assign ATA LUN's from a single raid group to both SP's then drive I/O's to both LUN's, then the drives ownership has to be passed around, and this is why you see very poor performance in those situations.

This "feature" is not code dependant, it's just ATA. One thing to note is that the LC-FC drives are dual-ported so you can spread the LUN ownership on those drives.

Regards,
Glen.

Thanks Glen,

I couldn't remember exactly why it was, but I knew the Best Practices docs were very clear on it. It does make a lot of sense based on your description.

I know we had done it wrong before finding out about this and were having some severe performance problems (and really regretting getting the ATA drives) until we figured it our. Now they work just fine (for the right application... usually just file servers).

man ..i am glad i subscribe to these boards ..now i have to do some trespassing to help my ATA users.

On the one hand I'm sorry to make work for you, but I'm glad you found out now instead of when the business areas were whining about performance and looking for heads to roll.

If they haven't noticed it yet you can still be a hero

strangely, i have not heard any complains..but then again ..these ATA luns are only used for windows file servers and sql dump drives. Actually the timing is really good. We are getting ready to upgrade flare, so before the upgrade i am going to change default SP owner appropriately, so when the SPs reboot, LUNs will be distributed correctly..at least i hope they will. I hate trespassing luns on running boxes, even with PowerPath.

i believe you need to use the java cli for that. java -jar archiveretrieve.jar

Hi all,

Thanks for your posts. Luckily, by accident, we will not need to do too many tresspassing, as most of the LUNS in a given raid group are associated with the same SP.

BTW, what command do I actually use to pull the nar report from navicli? I am hoping automate this every 24 hour using some sort of a batch script if possble. I have so far only retrieved the NAR file using the GUI from the navishere analyser tool.

thanks

I'm not sure (from your post) if you caught the whole point of the SP assignment issue, so I'll just clarify again.

The issue of assigning LUNs from the same RAID group to different SPs ONLY affects ATA drives. on FC drives you should be load balancing the LUN activity between SPs.

I apologize if I misunderstood your post, but I'd rather annoy someone and make sure they have the right info than have them happily misinformed:-)

Sorry Allen, I do understand. I should have emphasised that the issue was on the ATA LUNS. Thanks for your help on that.

However, I still struggling to get any reports(*.nar) using some sort of CLI to be able to automate the for regular archive.

When I use the naviseccli i get the following response even though the manual says i should be able to use status/archive switches ( I am using Navisphere NaviSECCli.exe Revision 6.19.0.4.14)

C:\Program Files\EMC\Navisphere CLI>NaviSECCli.exe -h x.x.x.x analyzer -status
Invalid Command. Invalid command line parameters.
usage:
analyzer
-set
<-narinterval seconds>
<-rtinterval seconds>
<-default>
-get
<-narinterval>
<-rtinterval>
<-duration>
-logging
-reset

Also when I run the Java version I get the following ---please help!!!!!

C:\Program Files\EMC\Navisphere CLI>java -jar archiveretrieve.jar -User X
-password Y -Address x.x.x.x -File archive.nar -Location c:\
temp -Overwrite 0 -Retry 2
Error returned from archive create:
Error returned from the Management Server on x.x.x.x:
Error sending request: Server and client could not negotiate the
desired level of security-versions are most likely incompatible-
Navisphere client software may be older than its server counterpart.

looks like you're referring to the manual of a newer cli release.
and the java cli error you get is quite clear, your array must be running a newer FLARE than the cli you are using.
so best to install R24 cli and try with that. both naviseccli and java cli can do what you need.

The best practice for changing ownership on LUNs is to change the ownership first, then use the host failover software (like PowerPath) to trespass the LUN - if you trespass the LUN on the array, the host will think it's a problem and trespass the LUN back.

Also on ATA disks, make sure you don't use ATA disks for Reserve LUN Pool LUNs.

glenk

regarding using ATA in snap pool. Is it a bad because you want your copy-on-write area to be as fast as the the LUNs you snapping/morror viewing ..etc ..so not to impact performance of the primary lun ? ...just trying to clarify for myself

Try this in a BAT file, then schedule the BAT to run every 24 hours if you have the archive interval set to 600 seconds. If you set the archive interval to 60 seconds, you'll need to run every 2.5 hours. Change the xxxx's to your systems and the filenme to what you want. I usually create a /archive in the c:\progran files\emc\navisphere cli\ directory - put the BAT file there and run it from there.

This works for release 14-19 - it should work for release 22 and 24, but I have not tried it yet - release 24 is "quite" different from release 19 - if you have release 19, be sure to read all the new Analyzer docs - new cli commands and a new GUI interface - also, with release 24, you can automatically set the array to collect ANR files.

glenk

For /f "tokens=2-4 delims=/ " %%a in ('date /t') do (set date=%%a-%%b-%%c)

For /f "tokens=1-2 delims=: " %%a in ('time /t') do (set time=%%a-%%b)

java -jar archiveretrieve.jar -User xxxx -Password xxxx -Scope 0 -Address xxx.xxx.xxx.xxx -File %date%-%time%-CX600-SPA.nar -Location "c:\program files\emc\navisphere cli\archive"

java -jar archiveretrieve.jar -User xxxx -Password xxxx -Scope 0 -Address xxx.xxx.xxx.xxx -File %date%-%time%-CX600-SPB.nar -Location "c:\program files\emc\navisphere cli\archive"