25 Jun 2013

Continuity and Cloud – Episode 3: Replication

Off-setting your back-up to a cloud provider either by Backup-as-a-Service or by utilising your existing software can provide a level of recovery for your business. Typically, this level of backup provides a recovery point of up to 24 hours and a recovery time of up to a few days.

With an increasing reliance on core computing services to keep a business operational, this is simply not sufficient to allow the business to continue running. Access to core data, particularly in times of busy trade can be a make or break point for many businesses.

Keeping data in synch between an on-site and off-site copy using either guest based, host based or hardware based technologies is a way to reduce the Recovery Point Objective down from the 24 hour backup-and-restore process. This level of data replication can also provide a greatly reduced Recovery Time Objective as well with your virtual or physical-to-virtual machines up and running in a very short time.

How do I get a better result?

Essentially there are three ways to achieve replication:

  1. Guest Based
    Using software to replicate the operating system writes or utilising in-operating system snapshots to copy point-in-time virtual machines
  2. Host Based
    Utilising hypervisor snapshot technologies to provide point-in-time copies of running systems.
  3. Hardware Based
    Using the advance features of your storage appliance or a separate physical appliance to replicate data from the storage network perspective.

Each one of these has recommendations and caveats. Guest-based replication can provide you with real time copies of the writes occurring in your virtual machine but can produce a tremendous performance hit due to duplication of writes. Snapshot technology provides you with great control of point-in-time copies, but has impact on the storage.

Before you jump into a replication solution, consider the following things carefully:

  1. Performance and Capacity
    Do you have the required spare performance and capacity in your current systems to support your replication solution?
  2. Network
    Does your current (or proposed) WAN solution provide you with enough bandwidth to get your replicated data off site in a timely manner?
  3. Do I really need to replicate everything?
    Replication comes at a cost, both in money and performance. Do you really need to replicate every workload you have? Can you protect some of your systems in other ways and reduce both the monetary and non-monetary costs of a replication solution.
  4. End-to-end control
    Does your replication solution provide you end-to-end control of the data flow?