M7 - Lesson 1: Defining levels of availability (What is high availability,…
M7 - Lesson 1: Defining levels of availability
What is high availability
Data center infrastructure:
The room that stores the server must have sufficient power and cooling capacity, and that capacity also must be highly available. You can make power highly available by ensuring that an alternate power source, such as a battery or a generator, is available when the electrical utility experiences outages.
For server hardware to be highly available, there must be redundant components. Redundant components can include power supplies, network adapters, processors, and memory. Error-correction code (ECC) memory helps to resolve minor memory errors.
To make storage highly available on a single server, you can use Redundant Array of Independent Disks (RAID). RAID uses parity information to ensure that a server can survive the loss of at least one hard drive, without losing any data. If multiple servers are available, you can replicate data between servers.
Ideally, you should use two different Internet service providers (ISPs) and two different physical connectivity methods. For example, one ISP could be land based, and the other wireless.
Active Directory Domain Services (AD DS) and Domain Name System (DNS) service are the two services that must be highly available to support the infrastructure services in the organizations. Multiple DCs and DNS servers
Within a LAN, this typically means redundant switches. Even moderately priced switches include redundant configurations. To make the network connectivity for any individual computer fault tolerant, you must configure redundant network interface cards on the computer.
Business impact analysis:
Business impact analysis determines an organization’s critical business processes and the potential damage or loss that can result from their disruption or failure.
Risk analysis identifies risks and their probability of occurrence. Risk analysis also identifies single points of failure, such as an organization’s disk drives, network switches, storage, or power supply.
SLAs for IT systems, both hardware and software.
Contact information and technical details of backup administrators.
A secondary site from which you can access critical applications and application data for critical business functions.
Maximum outage time allowed for your applications.
Network Load Balancing (NLB).
Failover clustering on physical or virtual machines.
Application-aware high availability.
Conventional data backups.
Virtual machine backups.
Creating a disaster recovery plan
Where should the recovered data be located?
When should the recovery occur?
What data should be recovered?
Select what data to recover:
In most circumstances, you will recover everything that was backed up. In some circumstances, you might choose to perform only a partial recovery to meet business-continuity goals, leaving a full recovery for a later point.
Choosing a data-recovery location:
It is possible to perform a temporary recovery to a Hyper-V host, and enable it to host the recovered server virtually until such time as the replacement hardware arrives.
Determining when to recover data:
waiting time for hardware components; partial recovery to alternate location or use DFS to create a new replica.
Testing the recovery plan:
recovery plan must be tested in order that you are sure that recovery steps are valid and successful. Failing to test the recovery plan might result in failure during the recovery process.
Evaluating the recovery plan:
Recovery plans must follow the change in an organizations IT infrastructure. If you do not update it, the recovery plan might become outdated and not correspond to the current configuration of the servers and applications in an organization.
Highly available networking
Redundant network adapters
Implementing MPIO with multiple host adapters provides you with alternate paths to your storage devices.
Redundant network equipment such as switches, firewalls etc.
Multiple WAN links / ISPs
Highly available storage
Highly available compute or hardware functions
Network Load Balancing