How about this FC connection(cascading) of VPLEX?

Hi hirach,

Many VPLEX customers do use a fabric for both front-end and back-end ports (redundant A/B fabrics required.)

There are a few more things to be careful for in your example on the t see VPLEX front-end ports (not world ending or anything like that, but just not recommended and will result in unexpected and not valid arrays and initiators showing up.)

So I think anyways you are just fine with the diagram on the right.  I would though be very aware of paths that requests take, so that you aren't say unintentionally traversing one or more ISLs if you don't have to.

HTH,

Gary

Hi Gary,

Thank you for your response.

This figure (on the right) is  just showing a drawing simply a part of our evaluation environment. We understood that VPLEX with the cascading connection is possible when setting FC zoning very carefully as you can see in the image in right hand-side. (After the world ending, actually, there is a distinct possibility toward the dawn of a new evaluation...)

Additionally;

why all the FCSWs are connected by cascading in our evaluation environment, is that the number of FCSWs will be insufficient to setting zone the host/storage through FCSW connection.

For VPLEX that was newly added to this environment, we did set the zoning of VPLEX front-end and back-end very carefully. Actually, some host does not satisfy the following condition: "don't have a connection to the same storage through VPLEX and also around VPLEX". And, the setting was done carefully so that storage LUNs that have been mapped to the VPLEX end-port will not be mapped to the other Initiator at the same time at least. In other words, we have done this setting to run "read/write" to a specific area via either one of "through VPLEX" or "around VPLEX".

Best regards,

hirach

> Actually, some host does not satisfy the following condition: "don't have a connection to the same storage through VPLEX and also around VPLEX". And, the setting was done carefully so that storage LUNs that have been mapped to the VPLEX end-port will not be mapped to the other Initiator at the same time at least. In other words, we have done this setting to run "read/write" to a specific area via either one of "through VPLEX" or "around VPLEX".

Ah yes, makes sense - Yes you can certainty have some volumes on the same host that are via VPLEX, and others via the storage-array.  If they're kept absolutely independent - sure, no problem.  Tread very carefully! :-) 

Perfectly okay to have the host zoned to a storage-array and VPLEX fronting the same storage-array (many wish we did the initial encapsulation this way so that there'd be zero host downtime to add VPLEX to the picture.)

Thanks,

Gary