Virtualisation has brought the cost of data centre- or machine-level replication down to the point where more organisations can afford it – but, it’s important not to assume replication can replace backups.
Replication and backup achieve two different goals. Replication is fundamentally about business continuity – if something happens that makes it impossible to use your IT infrastructure, having a live copy available in a separate location means you can keep your business up and running with minimal downtime. Backup, on the other hand, is about knowing that if any data is corrupted or deleted, you can still recover an earlier version.
The fundamental question asked by replication is, “how quickly can we get up and running again?” Backup answers the question “What data or systems can we get up and running again?”
This is what is expressed by the common Recovery Time Objective (RTO) and Recovery Point Objective (RPO) metrics. The important point is that they need to work together. You can have the most fantastic RTO that will get your organisation running again within seconds or minutes of a problem occurring at the primary site – but what if the problem corrupted or deleted all your data? If the problem was replicated along with everything else, you’re in trouble. Additional expensive software on top of Storage Based replication may be required as the only way to ensure you can always roll back to something usable.
In an ideal world of unlimited budgets, everyone would have everything backed up and replicated to within the last 60 seconds. But that gets very expensive, very fast – especially if you have a physical environment where replication means an identical copy of every piece of hardware and software. Virtualisation brings the costs down substantially because you don’t need identical hardware from servers to storage, but even so, replication is not cheap. Choices have to be made. This is where the concept of disaster recovery tiering comes into play.
Organisations need to survey all their systems and data to decide what level of priority to assign to each one. Tier 1 is your mission critical servers, apps and data – things that can’t be down for longer than five minutes, no matter what it costs to restore it. In most organisations that’s only about 10% of the total.
Tier 2, usually about 70% of the environment, consists of those parts of your system that are important, but can be down for as long day before it’s a real crisis. That means you can replicate and back up on a slightly slower and less expensive schedule.
Finally, about 20% of most organisations is Tier 3 stuff that only really needs to be backed up or replicated about once a week.
Having made these distinctions, CIOs and IT managers can set backup and replication schedules that meet all the organisation’s needs while staying within budget.
You may decide that replication is not essential given the cost. In that case, your backups can do the job of disaster recovery so long as you keep a copy off site at all times. If you have a virtualised environment, you can still restore the whole thing from your backups so long as you have something to restore to. That’s why it’s a good idea to sign up for a managed service that keeps spare capacity online and can make it available at short notice in case of a disaster.
By Warren Olivier, Territory Manager at Veeam Software