Disk Arrays
cancel
Showing results for 
Search instead for 
Did you mean: 

Best Practices for configuring CA

Balakrishna Kotian
Frequent Advisor

Best Practices for configuring CA

Hi Everyone, Does any one is having white papers on best practices for configuring CA.
Or what are the best points to be considered before configuring CA on new eva 6000(2no's).
I have found best practices for EVA with vcs 3.020 which is for EVA3/5K. I believe EVA 6000 comes with XCS 5.020 and above.

Tnx a lot for your time n effort.
13 REPLIES
Steven Clementi
Honored Contributor

Re: Best Practices for configuring CA

Bala:

Some things have changed with the new EVA's. Most of the rules are the same between the 2 sets of models, with some exceptions.


On the EVA 3k/5k, CA best practice was to configure Controller A Port 1 and Controller B Port 2 on the same fabric. With the 4/6/8k's, it is now... "Keep the even ports on one fabric and the odd ports on the other fabric". In other words, Controller A and B port 1 go to Fabric A, and port 2 on both go to Fabric B.

Additionally, I am not sure if this was the case with the 3/5k's, but with the new... your controllers have a mode... Master / Slace and there was a document that I received about it, but am not sure how exactly it plays into the picture of CA. I think it has to do with how the 2 EVA's replicate. I will have to find it and read up on it.

The new EVA's start out at 5.020 and there is a new release already (5.030).



Steven
Steven Clementi
HP Master ASE, Storage and Clustering
MCSE (NT 4.0, W2K, W2K3)
VCP (ESX2, Vi3, vSphere4, vSphere5)
RHCE
NPP3 (Nutanix Platform Professional)
Balakrishna Kotian
Frequent Advisor

Re: Best Practices for configuring CA

Hi Steven,

I have no idea on new EVA's and what major changes is done on new EVA as compared to EVA3/5k. Would you mind sharing the master/slave doc with me as I wanted to know more about the new EVA's.

Apart from placing even and odd controller ports on fabric switch is there anything more to do as per best practice.

TNX
Steven Clementi
Honored Contributor

Re: Best Practices for configuring CA

I will have to see if the doc is under NDA or not. Will let you know.


Steven
Steven Clementi
HP Master ASE, Storage and Clustering
MCSE (NT 4.0, W2K, W2K3)
VCP (ESX2, Vi3, vSphere4, vSphere5)
RHCE
NPP3 (Nutanix Platform Professional)
Steven Clementi
Honored Contributor

Re: Best Practices for configuring CA

Also, from what I remember, it was just a cabling doc on how to cable different scenarios based on all the existing EVA's.

E.G..
3k and a 5k
3k and a 4k
3k and a 8k
5k and a 6k
etc.

Steven
Steven Clementi
HP Master ASE, Storage and Clustering
MCSE (NT 4.0, W2K, W2K3)
VCP (ESX2, Vi3, vSphere4, vSphere5)
RHCE
NPP3 (Nutanix Platform Professional)
Balakrishna Kotian
Frequent Advisor

Re: Best Practices for configuring CA

About the cabling it is specified in CA planning guide.

I also got another question on positioning harddisk in enclosure. I have been taught in eva training that we start inserting disk from middle of enclosure for better performance etc. But I'm not able to find this stuff written anywhere in the HP doc. What is your comment on this.

TNX
Steven Clementi
Honored Contributor

Re: Best Practices for configuring CA

I have seen disks "everywhere". I believe that the EVA, when configuring it's RSS groups, takes a drive from each side of the shelf if it needs to select 2 drives from one shelf....

Assuming you have a 2C4D with all shelves full.

You create a 8 disk DG initially and..
The RSS would consist of the first disk in each shelf and the LAST disk in each shelf (looking from left to right). I have not tested this, but from what I have seen I think it is right.


Knowing this, either starting from the middle or from the ends would help the controllers load balance the disks properly...I.E. Putting 8 disks in bays 7 and 8 on each shelf OR in bays 1 and 14 VS. putting them all in bays 1 and 2 (in a 2C4D).

Yep, I could be wrong. Maybe I will test out that theory. I have 2 new eva6k's not in use yet.


Steven
Steven Clementi
HP Master ASE, Storage and Clustering
MCSE (NT 4.0, W2K, W2K3)
VCP (ESX2, Vi3, vSphere4, vSphere5)
RHCE
NPP3 (Nutanix Platform Professional)
Balakrishna Kotian
Frequent Advisor

Re: Best Practices for configuring CA

Yes, I agree with you even I have checked it works the same way but not seen any doc where HP recomending to plug hdd from the middle of the self.

Uwe Zessin
Honored Contributor

Re: Best Practices for configuring CA

About Master/Slave:
the EVA management agent talks to the master controller - guess how I know?
On the first EVA-5000, about 3.5 years ago, I have connected only one port to the fabric, but I could not initialize the system :-(
The reason was that it happened to be the slave controller.
(Running with one cable was a hack only, because I was too eager to try out the new system and nobody had moved it to its final position)


Disk drive enclosures tend to be filled from the center, these days (just for shipping balance), but for the first 3 years, all systems we received were filled left-to-right. The EVA should not care - it properly assigns loop-ids, even if there are empty bays within drives. Hey, it even managed to properly distribute mirror-members on a 2C3D (I am glad the VCS V1 days are gone ;-)
.
SAKET_5
Honored Contributor

Re: Best Practices for configuring CA

Bala,

Master/Slave controller relationship by all means is there atleast with EVA5K - a number of paramters at controller boot time determine a controller (one HSV110 out of the pair) to become the Master and the remaining as the Slave. This is not a fixed role assignment.

Again, I am not sure if I should be putting much more information on this design aspect of the EVAs here but can I ask why do you need to know this? as in what problem do you have with your EVAs that require you to figure out the master/slave relationship of the HSV110 controllers? The master controller obviously dictates a lot of stuff such as the firmware version - i.e. if one of your HSV110 controller fails and you replace it with a new controller that happens to run a later version of the firmware than on the existing operational controller (current master) - the master controller will downgrade the firmware on your newly replaced controller to bring it to its same version, and a few other things along these lines...

if you know how to access the fieldservice page from the SMA - you can determine at any point in time which controller has the Master role and which one has the Slave. Not that I can think of any reason why you should have to find this out!

an internal HP procedure to perform offline disk firmware upgrade procedure even requires you to determine the controller role and modify a certain value in case it is not set to the expected value - but this step is not listed anywhere in the publicly available HP document in updating your EVA procedure.

Hope, it helps.
Balakrishna Kotian
Frequent Advisor

Re: Best Practices for configuring CA

Hi Saket,

Thank you for the information. Currently I dont have any issue as I was planning to implement CA on EVA's. I wanted to know all the best practices required b4 implementing CA as I was not finding any documents related to CA best practices.

If you know any best practices which needs to be considered would be appreciated.

Uwe Zessin
Honored Contributor

Re: Best Practices for configuring CA

These days, you can find out the role of a controller (master/ slave) through the OPC, too. (Just don't select shutdown or uninitialize ...)
.
SAKET_5
Honored Contributor

Re: Best Practices for configuring CA

Bala,

I have already listed a few of the best practices and gotchas around CA deployment in your other post.

let me know if you need assistance with CA setup and i will be willing to assist.
Balakrishna Kotian
Frequent Advisor

Re: Best Practices for configuring CA

Hi Saket,

Thank you for your reply. Yes I would like to know how to configure CA, would be of great help if you have some kind of documentation. Currently I'm referning to the documents available on the website.