Ask a line of business manager about disaster recovery, specifically what data they need to recover and when by, and the answer is likely to be: “All of it – and now.”
As businesses depend increasingly on data to function, how to recover from an IT systems failure becomes increasingly important. Businesses and their customers are less and less tolerant of downtime, and of data loss.
So, IT departments are being forced to look again at how quickly, and how comprehensively, the organisation can get back on its feet.
In practice, this means organisations backup more data, more frequently, and need to restore it more swiftly. The RPOs and RTOs agreed with the business in turn determine the type of technology selected for recovery and business continuity.
As Phil Goodwin of analysts IDC points out, the trend is for businesses to move towards high availability rather than disaster recovery.
Developments in virtualisation and cloud computing have made that goal more realistic for a wider range of organisations.
Synchronously-mirrored data centres
The gold standard of business continuity and high availability is synchronous mirroring. For effective disaster recovery, the mirror needs to be offsite in the business’s own secondary or backup datacentre. Or, at a colocation site or a disaster recovery supplier’s location, such as 4SL or Sungard AS.
The location will depend on the risks and threats the organisation faces. Physical threats, such as extreme weather or terrorism, will mean the secondary site will need to be further away. But this puts more pressure on network infrastructure, and invariably increases costs. Companies will need high bandwidth, and possibly redundant, links between the two mirrors.
“This is the standard for business-critical applications and services, but it’s not cheap because you need two infrastructures and at least one robust, appropriately-sized link between them,” says Barnaby Mote, CEO at managed service provider 4SL.
Synchronous mirroring is the favoured option in industries that have a very low tolerance for downtime and very short recovery time objectives. These include financial services, as well as some areas of government.
Businesses often choose to manage synchronously mirrored datacentres in house, because primary and backup sites must be kept tightly aligned. This pushes up costs and reduces flexibility, as the technology platforms have to stay in sync.
Some businesses opt for a lower performance setup at their secondary sites in order to save money, but this will depend on how performance-sensitive applications are.
IT teams could also save money by only deploying synchronous mirroring for their most critical applications. In practice, organisations tend to lift and replicate entire environments, including all storage and data. This is because of the work needed to separate out critical applications, and the risk that by dividing up the infrastructure something would be missed and cause the copy to fail.
Failing over to a virtual environment (V2V and P2V)
Physical-to-virtual failover is a lower cost and potentially more flexible way of providing real time or near real time backup, and quick restoration.
The most obvious use case is for environments that are already virtualised. Companies have the choice to run in-house tools to replicate VMs, or use a service provider. VMWare’s vSphere replication offers single and multi-site data protection, as does Microsoft’s System Center Data Protection Manager.
One advantage of failing over to a virtual environment is to be able to use shared infrastructure for the backup site, to cut costs and reduce management overheads.
Another trend is for suppliers to offer platform-agnostic VM replication services. This allows firms to run backups on alternative infrastructure, and to failover a heterogeneous system to a single backup platform. This also opens up the option to use the cloud for replication.
The model is best suited to organisations already running virtualised environments. Technologies exist for physical to virtual recovery (P2V) – restoring an entire backup environment to a virtual machine – but backup copies need to be created regularly and moved offsite. Nor is failing over to a virtual environment suitable for companies that need continuous access to their data, such as those trading in financial markets.
On the plus side, most suppliers now provide bare metal recovery for virtual backups, which will help bring the business back online quickly. IT can also store local backups of VMs for quick recovery – for example, if local hardware fails – at the same time as staging copies to offsite storage for true DR.
The choice of dedicated, or shared, failover environments will again depend on the business’s RPO/RTO requirements and its budgets.
And, while backup to a virtual environment is a good option to save costs and cut complexity, for businesses that run highly-virtualised production environments it can be the only practical option.
Conventional backup and recovery technology does not fit well with virtualised systems because of their shared infrastructure and especially, shared storage. A VM-specific backup system is the best way to avoid bottlenecks and ensure safe recovery of the VMs.
Failing over to the cloud
The cloud really extends options for businesses to backup their data, and their production environments.
Suppliers that offer backup services for VMs increasingly offer cloud storage too. Acronis, for example, provides platform-agnostic VM backup, while Microsoft Azure can back up Azure VMs, SQL and business’s local VMWare machines. Products from suppliers such as Veeam and Commvault also support replication to the cloud.
Businesses can save money by only spinning up virtual servers when they invoke their DR plan, but they will of course have to pay for storage.
“Recovery into the cloud is feasible primarily for virtual environments; physical machines are problematic,” says 4SL’s Mote. “A range of RTO/RPOs are possible depending on the technology.”
Cloud services also work well for smaller businesses that might not have the skills and staff to run backup hardware. Companies can start with simple, online file storage or lower-end data backup services and scale up to application or VM backup as they grow.
“The IT used to assist in DR has matured greatly over the course of the past decade,” says Freeform Dynamic’s Tony Lock.
“Options exist at hardware level to copy or snapshot and replicate data between similar platforms, sometimes even over distances that previously would have been either impossible or prohibitively expensive. At the same time, software tools have been developed to deliver similar capabilities but between different hardware platforms.”
Backing up to hybrid environments is more complex than either straightforward physical-to-physical backup, or replication of an entire (virtualised) environment to the cloud.
On paper, a hybrid approach allows the business to decide which workloads should be mirrored, backed up to a virtual environment, or replicated to the cloud. The challenge is to decide which data and which workloads go where, and to maintain consistency.
Recovery from a hybrid environment will also be more complex. IT teams need to ensure all elements can be recovered from each platform after a full-scale disaster. But they also need a plan to deal with operational recovery, such as deleted files. Data egress charges levied by cloud providers are a cost that is easy to overlook.