Issue link: https://epubs.iltanet.org/i/4636
www.iltanet.org Infrastructure Technologies 9 revisiting disaster recovery in a virtualized environment There was complexity in maintaining the backup • schedules on each server The CPU load created on each server during • backups affected the user experience There was significant network traffic during backup • Tape backups were slow to restore • Tape media was expensive • • rsync/DFSR rsync was used to replicate files on the Netware servers while DFSR was used for the Autonomy iManage documents on the Windows server, as the Windows port of rsync did not replicate Windows New Technology File System (NTFS) file permissions correctly. The weakness of both these tools is the inability to backup open databases or files, and the degradation of performance as the number of files to be replicated increased. • VRanger and vReplicator This software pairing can either run on a stand-alone machine connected directly to the storage for SAN speed data access, or on a VM accessing data at LAN speeds via VMware Consolidated Backup (VCB), which utilizes the ESX server to access the SAN. At the time of backup, the product triggers the virtual center (vCenter) to create a snapshot of the VM to be backed up. When utilized with VMTools, which integrates with Microsoft's Volume Snapshot Services (VSS) in each guest, it quiesces databases such as SQL, Exchange and Windows Active Directory to ensure an equivalent powered-off backup of the VM. Subsequent backups are incremental, which yields a significant space savings. The resulting monolithic files are saved to disk for backup to tape or replicated to a DR/standby ESX server. Restores can occur at the VM or file level and are significantly faster than restoring from tape. Organizations considering vRanger and vReplicator should review Veeam's Backup & Replication product as well. • Double-Take This product was never implemented at Davis because of the change in DR technology at the firm. However, this was on the roadmap for use as it performs server replication and failover, which includes the necessary changes to IP addresses, Active Directory, etc. for the DR copy to be fully operational without user intervention. This technology was planned for core servers in order to minimize failover time and resources. BENEFITS This system, which used a potpourri of tools to achieve the end result, was cost-effective because it utilized existing or free tools from various vendors and allowed time to baseline the systems to better understand the long term requirements of a DR system and avoid expensive SAN replication technology. CHALLENGES As we expanded our virtual environment, these issues came to light: • It turned out to be difficult for an IT person and impossible for a layperson to activate the DR site • Keeping cutover documentation up to date was a challenge