High availablity suggestions

Simple question:
If storage is in the failing system how would you reach that storage

So the answer MOST PROBABLY is Yes it needs to be independant from the database engines.  And it will need to be HA enabled as well.!!!
Otherwise you have created a HA fronend in front of a SPOF.

You might try if Shared SCSI bus is possible.. (SCSI can do it, but not a lot of OS's can cleanly handle this, i think only OpenVMS is correctly equiped, and possibly Linux + GFS2 filesystem).

Shall we use any other storage disk other then San storage?
I mean to confirm whether we can use network share path

You do not need shared storage cluster for high availability with SQL, you can use log shipping. Basically you have a second server with a copy of the database and every transaction gets sent to the second server to keep it up to date. A witness is not required but can be added to provide automatic failover. https://docs.microsoft.com/en-us/sql/database-engine/database-mirroring/database-mirroring-and-log-shipping-sql-server?view=sql-server-2017

Traditional shared storage clustering also works, host connectivity can be fibre channel, iSCSI or SAS. When using SAS make sure the storage has intelligent controllers rather than dumb SAS interface modules that do the RAID and mirror each other's cache - they cost a similar amount to iSCSI rather than half the cost (which would be a dumb shelf).

My suggestions; Build your current database architecture consists of two nodes supporting application cluster to provide DB high availability, and scalability as well. To accomplish a fully high availability solution, have a disaster recovery site(s) is/are configured to support primary site failure. The technology used to perform data replication between these sites is “SAN to SAN” replication.

Is San storage is really required to use Availablity on Failover clustering?

Yes, if you're going to use a SQL Server failover clustered instance.
If you want to avoid the usage of a SAN storage then you need to check for other High Availability options.

Is Witness server is required with database data or it can be any listener static ip?

A witness server only exists for a Mirroring solution and isn't even a demand.
Listener only exists for an AlwaysOn solution so I can't see why are you mixing these solutions.

@madunix - i hope just for logfiles SAN->SAN replication... if database is copied that way it might have incomplete transactions & logs not yet fitting with those. Shipping logs & replaying those at the secondary server is that only safe way for transactions.

You can do it via SAN replication, In the past I did multiple implementations (synchronous and asynchronous replication).
https://www.experts-exchange.com/articles/17534/Storage-Replication.html

Synchronous replication ensures that every single write that occurs at the production site will be carried out successfully at the recovery site  Due to the type of business OLTP, a key need is to minimize the amount of data lost during a disaster. Synchronous will ensure that data at the recovery site stays in complete synchronization with data at the production site.  

Lately I implemented stretch cluster; With Stretched Cluster, the two nodes in an I/O Group are separated by distance between two locations. A copy of the volume is stored at each location. This configuration means that you can lose either the SAN or power at one location and access to the disks remains available at the alternate location. The SAN Volume Controller keeps both copies of the storage in synchronization, and the cache is mirrored between both nodes. Therefore, the loss of one location causes no disruption to the alternate location. This Solution achieves high availability on the storage level with active-active setup , For High Availability at the hosts/application level, this requires clustering software at the application and server layer to failover to a server at the alternate location and resume access to the disks.

If the active site gets severed from the network connection the other one will miss information. .. how bad that is depends on what is missing.
Logs are the most safe in this regard.

It is only with synchronous replication achieve zero data loss. With asynchronous replication, the application receives the feedback already, if the data resides on the local disk. The transfer to the secondary system takes place after the event on the background. Data loss can therefore not be completely ruled out. A significant disadvantage of synchronous replications the delay of the feedback. Which depends, among other things, the distance between the mirrors. One kilometer of fiber optic transmission delay as the signal by about 5 microseconds. I do agree with you both (good infrastructure and redundancy high speed links) must be addressed to achieve high availability.

@bhuva nesh please respond to the experts otherwise close your question.