I used vSphere 6. HA cluster works well. But i want to 100% up time to VM. (after one ESXI server fails VMs are restart on other ESXI)
Is there any possibility to configure HA cluster without restarting Virtual Machines.
I think you're confused here - ALL that HA does is restart VMs in the event of host loss. If you need a VM to survive a host failure without incident you need to use VMWare Fault Tolerance (FT). This is a licenced feature that requires at least 1Gbps of dedicated network connectivity between applicable hosts, most people use 10Gbps for this. With vSphere 6 you can 'FT' VMs with up to 4 x vCPU, this was limited to 1 x vCPU with previous versions.
The VMware Fault Tolerance feature is the option that will keep a VM running in lock-step on two hosts. High-Availability just reduces downtime, but still reboots the virtual machines.
However, these days, physical servers just don't fail, or at least shouldn't. What can you do to make your host servers more resilient? What was the nature of the failure?
Not to blow my own trumpet but I previously worked on the VMware High Availability/Disaster Recovery (Business continuity) architecture.
Previously the ability to monitor both the host software (e.g. Windows/Linux) and also application level software (e.g. SQL Server/Exchange etc.) was integrated into the same application.
This has been spread out between different applications now within vSphere. Basically the High Availability solution that is offered today only provides the user with the knowledge that if a VM is in trouble it will be restarted and dealt with without any user intervention. Handy for not phoning the admin at 3am but overall doesn't quite give critical application availability.
On the other hand we have Continuous Availability, now handled by Fault Tolerance, with this the state of the host software and application software (if required) will be monitored and when an issue occurs the system will automatically
failover to a secondary VM that will be (as close as possible) a replication of the primary VM. This provides you with near enough 100% up-time (we're talking less than a minute to switch-over, basically seamless to the user).
If you require 24/7 up-time I'd recommend that you look into Fault Tolerance to give you continuous availability.