Couple of points to add to Paul’s answer:
· Gatekeepers are still required
· SRDF, both Sync and Asynch, has been enhanced. The main focus is improving performance of both technologies. When initially release we won’t be able to replicate between a DMX and a VMAX3 but it is something we are looking at for a future release
Thank you for this ATE session.
- Wanted to know how the integration of the new VMAX is, w.r.t VPlex. What are the enhancements that have been added?
- Do the engines on the new VMAXes also share a processor? How are they configured?
It had previously been a nightmare when we had more than 35-40 hosts crowding 2/4 FA ports on each fabric, to avoid the other 2/4 set of ports since they were connected to hosts with more I/O loads. This was on VMAX 40k as well as VMAX 20k.
Also, this might be trivial (but maybe not), do the new VMAXs with the new code family - do they also require the same Director / FA settings when integrating / connecting new hosts?
Example: Setting the 'D' bit for HP-UX, SCSI3, SPC2 etc settings for Linux VMs as RDMs from ESX end.
Are these still the same with the new VMAX3?
Here’s the answers to your questions:
· In terms of VPLEX there aren’t really any new features day one but stay tuned for future enhancements
· Each VMAX3 array has dual socketed director boards with varying # of cores per Engine:
o VMAX 100K: 24 cores/engine
o VMAX 200K: 32 cores/engine
o VMAX 400K: 48 cores/engine
· These cores are allocated to FE and BE during platform sizing and pre-configuration. At GA there are three supported workloads: Balanced, Front End heavy and Back End heavy depending on the specific customer use cases
· For FE support for instance (as outlined earlier in this thread), a pool of cores is allocated to each emulation type (e.g. if FE ports are all FC then the FE’s are made up of one large pool of cores dedicated to service all the FC ports).
Does this answer all of your questions? Please let us know if you need more.
Thank you Tom. This is helpful.
I see that the dispersion limit still remains the same as with the previous models. Not necessarily a deal breaker for many, but, just curious - are there plans to increase this so that it facilitates movement of engines/ bays inside a D/C?
There are no plans to increase the 25 meter dispersion limit, but remember that now every engine (or a cluster of engines) can be individually be dispersed 25 meters from Engine 1. This provides very good DC flexibility.