Highlighted
owingj1
1 Copper

Brainstorm a Solution / Architecture

Really shooting for everything from the most basic to the most creative ways to solve this issue.

There is a customer that has mobile locations that are fairly large and contain 2 datacenters. Let's say it is a starship (to catch interest of our geekiest readers). One datacenter is way up by the guy in the big fancy command chair that now does priceline commercials. The other is back with the Scottish engineer.

The application running in these datacenters is an app that needs to be online at all times because it sort of prints money for the business (think of a city in the middle of the desert with bright lights).

How can we achieve the highest uptime? Automate failover in case of a "torpedo mishap" in one of the rooms. Currently there is strong preference for IP protocols (iSCSI). Adding a seperate Fiber Channel network may not be cost effective.

What currently available solutions are there from EMC and VMware? Without cutting into the profits. It can be more expensive but we would need to justify the extra cost with how much better reliability/availability would be vs the currently working iSCSI solution.

0 Kudos
3 Replies
orcsab
1 Nickel

Re: Brainstorm a Solution / Architecture

VMware Fault Tolerance remains the king of uptime.  Zero data loss in the event of a server failure for all applications.  This is useful if a torpedo hits the saucer, one of those sparks goes off in the background, and Ensign Yellowshirt drops to the ground dead next to his smoking panel.  Its limitation is that it only runs on single vCPU VMs.  Also the network load between the two VMs varies based on workload IO.  It can be significant.

VPLEX is great for active/active use of each datacenter to avoid downtime.  Should the crew need to centralize on the bridge to turn off life support in the rest of the ship this may be useful.  But, I believe you have excluded this for your "no fibre" requirement.  Shame, though.  Really incrediable availability if this is present.

Replication with continuous data protection (like RecoverPoint) can guarantee a zero RPO recovery in the event of a saucer detachment.

Scott

owingj1
1 Copper

Re: Brainstorm a Solution / Architecture

Thanks Scott.

I am wondering if there will be a NAS solution coming soon.

0 Kudos
orcsab
1 Nickel

Re: Brainstorm a Solution / Architecture

Yeah, I can imagine. 

Officially: this is a very interesting idea. It is something we are discussing internally.  EMC will not commit to bring this feature to market nor provide an estimated release date.

Scott

0 Kudos