managing vmax3 all flash mainframe only array

from what i know All flash models (F models) do not yet support mainframe (CKD), whereas hybrid model do support CKD.

managing dedicated vmax3 storage array is similar to VMAX40K from z/OS level, whereby provision CKD GK devices through FICON channels.  if managing through opensystem, then through FC present GK (FBA) devices to management host and install symcli and unisphere from opensystem management host.

optionally though, at initial discussion - you can ask for embedded unisphere which will be hosted on VMAX3 itself - you just manage using browser poininting to IP assigned embedded unisphere

you can managed multiple vmax3 from external management host by provisioning FBA GK devices.

Thanks for your comments .....the original AF models didn't support CKD and FBA on same unit ...which is why we have now one new 450 AF  for each task, z/os and open systems.

However, currently (May 2017) i understand that CKD and FBA can now coexist on a new AF array...as of course they could (and do here) on a VMAX40K.

With hindsight it would have been helpful if our client had been able to purchase TWO rather than FOUR arrays !!

Our wish , as we do presently on  our hybrid mainframe & open systems VMAX40K, is to have ability to manage CKD disk groups via a FBA unix host, hence the question over the apparent dilemma of the absence of any  potential FBA GKs on our new dedicated z/os array, which, because of its use, doesn't have any FBA allocation (at present).

So question is whether we can carve out some FBA GKs in some way on our mainframe array. Does anyone know if this possible ?

https://community.emc.com/people/PaulCork/blog/2016/03/11/embedded-management-for-vmax-all-flash-and-vmax3

review this link as well.

Yes thanks SKT …had already read this useful article particularly re ability to manage eg a z/os array via eMGT , connecting over IP to symcli (unix) host. My remaining question is whether able to use symcli from open systems host by in-band means.

We appear to have a non-typical environment too , since our z/os and open systems storage will be on separate 450 AF arrays, rather than on the hybrid array we are used to managing (from open systems), since, at point of purchase, hybrid usage was precluded on AF.

do you have regular FA, or FICON only adapters ?

thanks for the update on AF support for CKD.

yes, we can create FBA GK devices,  VMAX3 hypermax os needs FBA for internal space for containers/VMs - therefore while sizing some % space dedicated for FBA. so there must be a narrow FBA space.

you can create FBA GK devices and  present to opensystem management thorugh FC ports. since we do not expect anything to be written to GK device we can turn these gk vols as WD, just to be safe with this narrow FBA space on a dedicated CKD SRP; as you'd know in VMAX2 we can present FBA GK devices without binding, but VMAX3 automatically binds.

you can create thin GK, they don't need to be bound to any SRPs

hi dynamox,  to your earlier question ...yes we have FAs in slot 2 as well as our FICON modules....i'm not yet up to speed on concept of SRPs (but have administered VMAX, VMAX40K).... our requirement is exactly as you put it ...to be able to create thin GKs (somewhere) we can use on our existing symcli hosts. My concern was that all of the available storage would be assigned to a single disk group which in turn would comprise the CKD pool, ie not afford any chance for any FBA GKs.

see if you can create a gatekeeper device. If you can go ahead and create a masking view and so what you did with your other arrays.

symconfigure -sid 1234 -cmd "create gatekeeper count=1,emulation=fba,type=thin;" commit

John,

Do they already have embedded management on the VMAX3 you are asking about?  If so, they do have a small amount of FBA already defined on the system.