I made an unspoken promise to myself that I'd never blog a stereotypically technie post, but it seems like such a waste to write so much every day and not show any of it around. Given below is a random snippet I was forced to write in half an hour for a tender that closes tomorrow. Sheesh, as if life wasn't hard enough.
Citrix Disaster Recovery
Based on business requirements and practical implications, there are two recovery models to select from:
1. active-passive
2. active-active
The active-passive model is one in which the disaster recovery data center is in a warm-standby mode until required. All users connect to one of the data centers. The remaining data center is fully redundant, and remains unused until a failure occurs.
Active-active describes a disaster recovery environment in which the site designated as the disaster recovery data center is online and functions in conjunction with the primary data center. In this model, users connect locally to the site that has the least latency.
Apart from Citrix, provisions must be made to ensure that the following infrastructure is in place:
• Physical network infrastructure (routers, switches, and so on)
• Directory Services (Active Directory/Novell eDirectory/LDAP)
• Network services (DNS, DHCP, and so on)
• Data storage and replication
There are two aspects with regard to Disaster Recovery planning in Citrix:
1. Redundancy of the physical server components. These are:
• Redundant power supplies
• Fault tolerant RAID setup depending on business requirements (for example: RAID 1, 5, 1+0)
• Fault tolerant network interface card (NIC) teaming
2. The redundancy of the services that the physical servers provides.
e.g.-
• Data Store
• Data Collector
• Farm Metric Server
• Database Connection Server
• License Server
However, with regard to smaller Citrix infrastructure deployments of about 4 servers, it is unlikely that there are dedicated servers performing these functions. As such, it is recommended that a fully redundant solution is implemented in active-active configuration, as it will also resolve any irregular performance issues that might possibly arise during peak utilisation.
The following procedure could be used to perform this:
1. Replicate the existing infrastructure on to 4 other servers
2. Configure all 8 servers as part of one Citrix Farm
3. Create two zones within this farm, the first zone with the current infrastructure, the second with the new servers
4. Define a zone preference and failover policy, filtered by user, IP address range or other appropriate criteria
5. Based on the connection method, configure clients to accomodate the different zones
