Moving, ed.harnish@acronis.com Ed Acronis Harnish, Inc. VP Managing, Maintaining Data in a Virtual World BR The VM, Console DR and & Archiving and Recovery Operating Recovery Strategies System Agenda High Disaster availability Recovery within Considerations the datacenter Some Definitions 1-2 weeks 1-2 Months Ongoing Data / Server Data Server B/R D/R Archive Some companies use the same procedures for BR / DR 1
ESX Server ESX Server VMKernal Configuration files Base OS Hardware Interface Treat ESX Server and VMs as separate resources with different / Recovery requirements Backing up Service Console Server 2
Over the Network Server Easy and inexpensive to implement Fits with many data center standards Best design for Multiple server environments Does not increase load on the Service Console Does not work well for remotely located ESX Servers Connected to Local Device Simple to use and implement May work well for remote location ESX servers Hard to manage in large environments Local Console software may not be able to backup VMs Will increase load on the Service Console Device assigned to a VM Removes load from the Service Console Cannot restore a host without first recovering a VM 3
Critical components of the Recover faster /etc/profile /etc/ssh/sshd_config /etc/pam.d/system_auth /etc/ntp /etc/ntp.conf (optional) VMware ESX Configurations VM Configurations Backing up Service Console Server 1) 1) Over network 2) 2) From Console 3) 3) Through a a Virtual Machine (VM) 4
VM like a physical server VMX VMFS Etc. Allows for partial or file level restores Easily fits into most existing backup schemes Requires no scripting or VMware knowledge Does not allow for a VM to be recovered in whole Long Recovery times VM and Recovery Treating a VM like a Physical Server Network Server Agent Virtualization Layer Guest File System Server.vmdk at the VM level VMX VMFS Etc. Allows for a complete restore of a VM flexibility (Incremental, Full) Does not easily allow for file level restores Large amounts of data Long backup times May hit 2 GB limit of some backup software VMs must be stopped 5
at the VM level Network Server* Agent Agent Agent Agent #1 #1 #2 #2 #3 #3 #4 #4 ESX - - * Through VCB Clusters Data Center Raw Device Mapping Raw Disk Public VMFS Node1.vmdk Shared VMFS Shared.vmdk Public VMFS Node2.vmdk Cluster Node 11 Cluster Node 22 ESX 1 ESX 2 6
Clustering the ESX Server - Provides for automatic restart of VMs - Does not provide for service failover within a VM - Additional cost - Additional level of complexity The Steps 1.Drive 2.Power 3.Verify 4.Power Dream after a failure environment 5.Get to on secondary verify configuration facility of a datacenter of ESX hosts DR 6.Get coffee you up VMs can see the order of VMs DR plan the priority SAN Considerations raise for while getting watching all servers the VMs up power up and in under options an hour Primary Location SAN Replication Second Location SAN ESX Servers X2 VM Configuration Same IP Subnet in both locations 201.03.02.2 201.03.02.2 VirtualCenter The Dream DR - What does it take 7
Typical ESX Server Recovery A contract facility that will provide you with like hardware The repurposing of existing company owned hardware masked ESX host masked ESX host ESX host Prod Array masked ESX host One size does not fit all The VM Console and / Recovery Recovery Operating vs. Strategies Disaster SystemRecovery High availability and within Recovery the datacenter Summary Disaster One size Recovery does not fit Considerations all 8