Metadata

Metadata is information about a document, such as author, camera used and email attachments, which can be read without opening the document itself.

See my article http://www.filecats.co.uk/metadata-extended-document-properties/what-is-metadata/ for more details.

in a SDDC (software defined data center (hyperconvergence))  the metadata may consist of performance counters and using these performance counters you can scale up  the vm's to faster storage or a faster machine or scale down or  scale up/down instances of a vm.. consider that you have compute capacity with different levels of servers, storage capacity with different levels of performance. you would use this metadata of performance to change the compute and storage units on the fly as the performance reaches specific levels.  For Windows Servers/Clients in the SDDC you can also roll out updates to the machines in a prescribed cycle. i.e. we want to update server1 so we replicate server 1 onto a different host and when it is running we switch the networking so packets going to server 1 now go to server1 on host 2, and then on the original host we service the machine and once complete we seamlessly rotate the two vm's so that server1 on host2 can be updated or think of it as a load balanced situation with 3 machines as each machine is being updated it is removed from the pool and then once done and up and running added back to the pool so another machine can be updated. if the machines are being hammered and need more cpu power you can seamlessly replicate the machine to a faster more capable host and have the network switch the packet paths.  Same goes for storage where you have superfast ramdisks, fast SSD's, medium speed spinning HD's, and slower archival during a boot storm (envision start of the business day)  that the VDI machines are booting, the source disk can be moved/cached from spinning to fast SSD or SF RAMDISKS, now once the boot storm is over then the SSD's can be used for SQL servers or wherever the demand is.  All of this requires metadata to be collected to determine the loads and to balance the loads.  

Another item to consider is compute scaling for  the load balanced machines add/remove machines from the LB pool on demand.

Thanks