Remote and branch offices can be a major headache when it comes to backup. But it is no less important to back them up than it would be at your main datacentre.
Remote offices often run outdated backup infrastructure and are dependent on non-expert staff. So what are the key best practices for remote site backup?
Archana Venkatraman, datacentre editor at ComputerWeekly.com, talks to Antony Adshead, storage editor at ComputerWeekly.com, about the key characteristics of remote site backup and the main options available that comprise best practice, including the use of the cloud as a backup target.
Venkatraman: What is special about remote site backup and what are the main choices available?
Adshead: Obviously, different organisations’ remote office backup setups are going to be different, but there are some fundamentals that are likely to apply, and that have a bearing on remote site backup.
One key factor likely to be common to all is that there will be few or no IT staff on site, and that non-IT staff may be responsible for backup-related tasks.
Another is that the backup infrastructure at remote offices may be different to what exists at main offices.
It may be just different, perhaps a legacy of acquisition, where another company’s backup systems were inherited.
And, it may also be more legacy because branch offices, simply by the fact of being on the periphery, can sometimes fall victim to being an afterthought or be perceived as less important.
The truth is that remote site backup is as vital as it is at a main datacentre. The data almost certainly has business value and may well be subject to legal and regulatory compliance rules that specify conditions of retention, integrity and e-discovery.
So, we are generalising here, but the likely manifestation of these fundamentals is that your branch offices could well be not integrated into the overall backup regime, are possibly reliant on older technologies, such as tape, and are being operated by non-expert staff. The bottom line is that they are quite possibly unreliable and non-compliant.
What we want to achieve, in terms of best practice, is to arrive at common systems across the organisation, with backups at local offices that are managed centrally, or at least are easy to use, and that make the most of currently available technologies and bandwidth.
Venkatraman: What are the main options for remote office backup?
Adshead: At the most general level, the three main options for remote site backup are:
- Local backups, to either disk or tape, and with or without removal of backups from the premises;
- Backing up to the main datacentre via the wide area network (WAN);
- Backing up to the cloud.
Let’s deal with the first one. Backing up locally is likely to be the most legacy of all these scenarios. Typically, it will comprise tape backup nightly, perhaps with a regime of incremental backups and then weekly full backups. The least secure variant of such a scenario would see tapes left on site, but in many cases these will be removed to another location to give some basic disaster recovery protection.
This approach is potentially unsafe for a variety of reasons. It is very dependent on manual processes, carried out locally and often by non-expert staff. It is also a scenario that has been superceded technologically – in backup software, in bandwidth availability, and in ways of making the transfer of backup data more efficient, such as with data deduplication.
Moving on to the second set of options, this comprises backing up to the main datacentre over a WAN link. There may or may not be a local copy retained at the remote site; we’ll come to that.
This option entails use of a backup product that manages the movement of backup data across the wire to the main datacentre. It could be the local instance of a centralised backup product implementation; all the main backup software products allow remote management of backups from the centre.
For this scenario to be an option you need enough bandwidth between remote offices and the main datacentre. Having said that, bandwidth is often likely to be a constraint, but there are ways of dealing with that.
Source data deduplication, for example, can cut down the volume of data that needs to be sent across the wire by cutting out duplicate blocks at the remote office before firing the reduced data set off to the target. Global data deduplication functionality compares data from the remote source across multiple locations and ensures duplicate data chunks are not sent unnecessarily
There are also WAN optimisation products. These use data deduplication techniques, but also use traffic prioritisation, caching and compression techniques to lighten the load on the network.
A potential fly in the ointment with regard to backing up to the main datacentre is any likely impact on apps running there as backups from remote sites hit the network.
And also – and this is important – what are the implications of restore times from data backed up to the main location? Are backed up files likely to be so large as to pose difficulty when they need to be restored?
If that is the case then data can be held locally, perhaps staged to disk for a period, as well as being backed up over the wire.
This could be on any local disk, but several suppliers make backup appliances that can hold data locally as well as send backups offsite.
Another way of making life easier on the WAN is to use some form of replication or snapshotting, sending smaller chunks of data over the wire throughout the day rather than in a single backup window, but such techniques are outside the scope of this podcast.
Venkatraman: Finally, what about the cloud?
Adshead: At one level the cloud is just another backup target, so similar considerations apply as when backing up to the main datacentre. As with that scenario you will need to consider whether to retain some, probably the most recent, data locally in case restores are needed and they take too long to come back from the cloud. In that case you can hold data on local disk or in devices marketed as hybrid cloud products.
For larger organisations there is also the option of contracting for "raw" cloud space with the likes of Amazon S3 and using that as a target for your backups.
Be aware, however, that using the cloud potentially opens up new scenarios that deserve due diligence before proceeding. Namely, there could be issues of compliance if the cloud provider holds data in other countries or continents. Laws that do not apply here could apply there, and vice versa.