Solaris

For example, assuming you have 3 local zones running on your Solaris 10, if one fine day, your system hangs or panics for some reason, all your 3 zones will be down (affecting production, provided your users use the zones as different servers. that's basically transparent to them). If you need to patch your system, you need to halt your zones as well.

In a Non-Live environment, implementing zones is ok with me. Its not as critical as in a Production Env.

Having a server crashing might happen whether it has zones or not and the application(s) running on it would stop anyway.
Regardless of whether you are using zones or not, having a single point of failure in a production environment might be unacceptable depending on the service level agreement expected. The global zone is indeed a SPOF for the NGZ running on top of it but nothing prevents you to build an reliable architecture with zones just like you can do it the traditional way.
Moreover, moving zones is far much simpler than moving plain OSes and patching or upgrading a system doesn't require to halt the zones when using live upgrade. Only a planned reboot is required to finalize the operation.

I'm definitely recommending using Solaris zones in all environments, dev, staging, production or whatever. They help optimizing resource usage, reduce operational and environmental costs and can help reducing security risks.

Though I understand the above description are appropriate, just to add or concise / summarize, its technology used by sun for server virtualization. In term which decreases the data center operation cost. Assume if you’ve five different applications which go off from time to time, so you can’t host all of these applications on same system. The option u’ve is to host them on different physical system (before) which in term use to increase the cost of data center in term of power / manpower / cooling and many more. Now a days instead of buying 5 system, you can buy one with high capacity and on that you can host 5 different application in either complete isolation or partial isolation, where in the changes done in one environment ( ZONE ) do not have impact on other env / zone.

kumarmani's point is excellent. there are many many unrealized expenses when deploying a server for a single host. cooling (crac, chiller plant), cable runs (usually around $700 a drop in my dc), power runs (available circuits, pdu capacity, ups and generator capacity), white space... the list goes on as this is something i interface with everyday. and yes, depending on the business culture, SLA, etc, zones may not be option. however, they are a great option for smaller shops that can't afford three m5000s to host 3 separate hosts. zones are great options even for financial institutions that may not need a single dedicated server for a project. i do this all the time actually.

unfortunately i am not a vmware expect but i would assume that without an esx server and vmotion, vmware is bound by the same points of failure that zones currently hold.

I don't get what SLA disadvantage zones could have compared with traditional Solaris installations. High availability can certainly be achieved with Solaris zones too. Just have two zones providing the same service on separated servers with a load balancer/failover architecture, a shared redundant back-end and you are set.

Compared to traditional Solaris installations, the management of zones could get ugly. There's also the matter of federal regulations depending on the industry.

Do you know of any regulation objecting Solaris zones or is it just speculation ?
I would expect regulations to recommend using zones instead.