And the process itself is becoming more complex and hence expensive. Specialised products offer faster recovery but increase system requirements, and many features are rarely used. Content indexing, for example, is a great idea but is rarely implemented fully.
If you look into a Data Centre for a 1,000+ user organisation around 20% of the server/storage rack space will be used for backup equipment and that is just the live stuff. There will be older versions of storage/drives/tapes hanging around “just in case”. By streamlining and using newer tools/capabilities virtually all that space can be freed up.So do we really still need backups? Heresy perhaps, but let’s take a closer look.
There are a number of reasons why we might need to recover information – data loss or corruption (e.g. malware), a bad configuration update which needed to be backed out (patching), hardware failure, or for disaster recovery and business continuity planning. It could also be to provide legal proof of information being in a known state at a fixed time, long term information retention or archiving, or off-site storage. These can be grouped into three categories: recovery, archiving or retention.
In my view, the majority of these reasons for backup are no longer valid.
Much of this is due to cloud. It can address archiving and retention, and provides a better solution for DR.
- DR using just backup isn’t viable; it takes too long, and the success rate was minimal. Many public cloud solutions can stand up a live DR scenario much faster and at a lower cost, and cloud or other provider services can replicate data elsewhere to provide offsite storage. A purpose designed cloud backup product also lets you work in de-duplication from the ground up, with the old environment kept for emergencies (which will reduce quickly).
- Specialised archiving and legal proof of information products are available, but most use a proprietary format, leading to vendor lock-in. Cloud works well for minimal use, while local storage will be cheaper for large volumes of archived data which are accessed frequently. Products are increasingly available that cater to this market and are based in the cloud.
The area which needs more consideration, and in my view is most important to the business, is data recovery. However, the reasons this might be needed can be handled in other ways.
- Hardware failure should no longer be an issue with virtualisation and redundancy. If the worst happens, it’s covered by DR.
- Bad OS updates should no longer happen with proper change management and testing. However, proper roll-back plans are essential, and you still need a copy elsewhere, as system/configuration data changes more than is often thought, especially with the requirement to patch/update constantly for security reasons. This data needs to be consistent but can easily be controlled by a template or pre-determined configuration/set-up.
Data corruption is always a potential issue, but storage and hybrid solutions have snapshot/replica capabilities that are database aware. There are also products aimed at using these capabilities which provide near instantaneous recovery. Additionally, for unstructured data – primarily single files - methodologies built into the OS can be used for quick recovery, and for longer term only changes need to be recorded.
There are several provisos to using cloud storage for backup. Latency will occur with large amounts of data. In most cases this isn’t critical – if it is, cloud is not the solution for you. Storage costs rise with significant volumes of data but can be reduced by being clever in your choice of storage, while data egress costs are typically around £0.05 per GB for large amounts. If security is a concern, either choose a supplier who meets your concerns or do not use cloud.
If implemented correctly a significant amount of “tin” can be removed from your computer rooms and the typical historic convolution inherent in long standing backup solutions be discarded. This does require you to be brave and re-work retention policies on backups, to specifically be relevant to the important information, but there is so much waste in the current systems that it needs these swingeing measures to make a real difference.
I hope I’ve convinced you that you don’t need a single product to manage your backup requirements. A better alternative is a mix of small and simple products that make best use of the cloud and other capabilities that may already exist within your infrastructure. This means slightly more management web pages, but these have been simplified and so require less technical skill.
Finally, bear in mind that your organisation is still the data owner. Another organisation can be paid to deliver your backup service, with appropriate SLAs, but the data remains your responsibility.