VMware vCenter, Microsoft VMM Manager, Xen 管理和生产Cluster最佳实践

mqzlp

　　Seperate Management Cluster from Production Cluster.
　　  Provisioning a management cluster to host critical infrastructure  management components like vCenter, Active Directory, SAN management,  and so on. To separate from production cluster.
　　A management cluster is a dedicated set of hosts (per site)  configured as a VMware vSphere HA/DRS cluster. The management cluster is  used to provide resources for virtual machines that provide management  functions for the virtual infrastructure. This includes (not all may be  applicable to every site); active directory (or similar), DNS/DHCP,  general management server (for Network and SAN tools etc.), vCenter  server, database server for management databases, web client server,  single sign-on server, Site Recovery Manager; vSphere replication  server, vCenter operations manager (and other components of the Suite),  vCloud Director cell (and related components), vCloud Networking and  Security Manager, data protection servers (e.g. CommVault VSA).
　　   The primary purpose of the management cluster is to separate the  resources used to manage virtualisation workloads (i.e. any  non-management workload, so this includes dev etc.) from those  virtualisation workloads themselves. In the event of a site failure and  subsequent recovery, the management cluster is the first thing you bring  up. This means you can get the management functions back first and  faster.
　　   So not only should the compute resources be separated (i.e. a separate  cluster) but consideration should also be given to separating the  storage provisioned to the management cluster (from the production  storage) – in a similar manner to how you would not retain backup copies  on the same storage systems as production data. If separate storage is  provided then this may be a dedicated array, and consideration can also  be given to the vSphere storage appliance which shares local storage  across a cluster of hosts (this would need further discussion though as  there are some limitations with the VSA).
　　   A further validation for establishing a management cluster is that  amount of resources required to service the management components can  easily equate to the resources available from one or more hosts. If  these resources are provisioned from production clusters then this a)  leaves less resource available for production workloads and b) makes it  more difficult to do capacity planning for production workloads (i.e. as  more management capability is added – think of adding metering and  billing at a later date – then less resources are available).
　　   The management cluster should ideally be a three nodes cluster per site  (best practises), although it is common to relax the three node cluster  best practise and have a two node N+1 cluster where N is 1 (good  practise since we still have a Management Cluster). Ideally the same  virtual networking should be used across management clusters and  production clusters (distributed virtual switch yes, but Nexus 1000V  possibly not unless you are implementing the physical Nexus 1100  appliance).
　　   In summary, as the reliance on the virtual infrastructure continues to  rise (think how now business critical applications are being targeted  more frequently now), then increased resilience of the management  components becomes critical.