Data Protection

The IT market offers a huge array of data protection products and product features. With so many vendors each promoting their own approach it can be very difficult to cut through the hype and determine what you really need.

The types of data protection commonly used in larger enterprises are summarised in the diagram below.

Data Protection Methods

Data Protection Methods

Data protection methods

While most of the terms used above will probably be familiar, it's worth defining them completely to avoid any confusion.

RAID

Fully redundant disk storage subsystem, able to tolerate the failure of one or more drives without any loss of data.

Snapshot

A point-in-time static copy of a disk volume or filesystem. The snapshot may be created either in hardware (e.g. disk array controller) or software (e.g. server filesystem, disk management utility or the application itself.

Replication

Real time duplication of data to another storage subsystem, either at the same location or - more commonly - at a remote or "disaster recovery" site. Data is copied as it changes, so that the duplicate is never more than a few minutes "out of date".

Tape backup - on site

Data is backed up to tape and the tapes are then stored locally, either in the automated tape library itself or elsewhere in the building such as a fire-resistant safe.

Tape backup - off site

Data is backed up to tape and the tapes are removed to a secure, remote location such as a storage vault. To meet typical disaster recovery requirements, the remote vault needs to be at least 10 to 15 km away from the primary site

Thorough and effective data protection for an enterprise requires a combination of online and offline methods. The table below illustrates the effectiveness of different data protection methods for a range of data loss events.

Event Coverage

* only if the problem is discovered and fixed within the life of the snapshot (usually 24 hours or less)

While offsite tape backup clearly wins out from a coverage viewpoint, it has the drawback of much longer time to recover than many online methods. As a result, it's becoming common for large enterprises to use online methods as their primary form of data protection with tape backup as a supplement or fall-back approach.

Data protection planning is always best begun with business requirements, so kick things off by asking some questions that could form the basis of Service Level Targets (SLTs) for each of your core business applications. For example:

  • For this particular server or application, what does it cost the business while the data is inaccessible? Consider unplanned outages of several different lengths, depending upon the importance of the application, e.g. 1 minute, 1 hour, 8 hours, 1 day, 1 week etc.
  • If we lost all the recent data for this application from the last, what would it cost to re-enter or re-create that data?
  • What other ways would data loss cost your business money? e.g. If your business delivers a service to customers (e.g. telco or utility) and lost records about customer usage, you may be unable to bill them for their usage over that period and would suffer lost revenue as a result. Or industry regulators may impose fines upon your business for a failure to meet legal requirements for data protection.
  • Is this application adequately protected against all common causes of data loss (see table above)?
  • Is this application properly covered by a Disaster Recovery (DR) plan? How often and how thoroughly is the DR plan tested?

The answers will vary for different applications and types of data, depending on the level of importance of that application to your business. Sophisticated data protection is more expensive than basic measures, so it makes perfect sense to invest no more or less than is justified in each case.

Let's begin looking at the implementation of data protection methods by looking at the limitations of distributed tape backups - the most basic approach and one that many organisations reach through normal, incremental growth in servers and data. Then we'll explore more advanced approaches.

Backup

Backup LAN

Secure Backup

Centralised Backups

Search the Data Engines Site

Featured Content

Backup or Archive? An age old question - after almost 60 years of data storage and backup on electro-magnetic media, people are still confused as to what a "Backup" is and what an "Archive" is. See Tim's blog post explaining the difference. 


The ENSTOR team pull together the 'noughties' most influential technologies in this Blog Post from Tim Davoren