Data Backup & Redundancy

Introduction

Having a data backup in case of equipment failure is highly important, as data loss can lead to lost records, lost revenue, unforeseen costs and sometimes even terrible consequences. If you use any computer that you rely on to hold important data or to perform important operations on a day-to-day basis, you must make sure that you have adequate backup and redundancy in place, to keep your work running.

Isn't backup and redundancy the same thing?

Kind of, although there are some small differences. For example, having multiple hard drives to back up your data would be considered redundancy. However, redundancy may exist in other aspects of your system too, such as having multiple power supplies, spare memory, CPU or motherboards, additional equipment, etc.

In other words, you can get as many drives as possible and have very expensive and robust storage RAID controllers, however should the motherboard fail you will be faced with downtime while the board is being replaced. Sure your data will be intact, however the system may be down for weeks until the board is replaced.

Making proper assessment

To determine how much data backup as well as how much redundancy you need, you first need to determine which one of these 2 things is most important to you. First, if you only care about storing and backing up your data, and you don't mind an occasional downtime, then you do not need to invest in multiple components, backup power supplies or other solutions. You should focus all your attention to getting the best data backup set possible for the money you're willing to spend. This includes but it's not limited to things like the NAS (Network Attached Storage) Storage servers, RAID controllers, multiple hard drives, external hard drives, tape drives, etc. All these can be used for either scheduled backups, on the fly backups or any other random backups you want to do.

On the other hand, some systems don't necessarily generate any useful data to be saved, however people rely on them to run every day and provide adequate services. Having any downtime on these types of servers is likely to cause loss of revenue regardless of whether or not data is safely protected. In these situations, having secondary hardware usually helps to keep the unit going. For example, one can purchase redundant power supplies. Some cases can carry 2 and even 4 power supplies, all which can take over for one another in case any of them fails. If your server isn't getting any power, it cannot get any work done. Failed PSU can render it useless and having an immediate backup is very helpful. In most cases, failed PSU will announce itself, and some will even trigger alarms through appropriate console management software, so that the system admin is aware of the failure. It's also wise to invest into a server with multiple CPU configurations, where by 2 or even 4 CPUs can all work at the same time, or provide redundancy for one another. Users can remove or swap CPUs that have failed, thus reducing the downtime to 0 even when waiting on parts to be replaced. This same principle applies to most other parts. Not all cases will allow for multiple parts.

Now what happens when you have a machine that is both, required to run at all times and provide services, as well as provide plenty of data storage and backup. Well, anytime you try to build an all-in-one unit, you will always have to give up the best of all possibilities. Kind of similar to an all-season tire on a vehicle. It is not excellent in winter, and not excellent in summer. It's just kind of good all around. But, if you wanted best possible, you would have a dedicated winter set and a dedicated summer set. Same thing goes for computer systems. Unless you can have dedicated workload servers and dedicated storage servers, you could technically combine them into one unit, however you will lose some of the advantages of each. Depending on the environment that these servers are put in, they might actually work great for you, but the redundancy will decrease significantly. Companies spend millions of dollars on always introducing new backup systems and redundancy servers. Depending on the importance of your data, you can never have enough redundancy, just in case of serious disasters. For example, an office or a building fire can wipe out entire data centers if not backed up off site. However, if bad weather strikes and you get flooding in an area, so if your storage is only located in the building next door, it's also likely to get damaged. As you can see, you can chase the redundancy issues forever, it all depends how far you need to go to protect your data.

Cloud storage

A new method of storage that has been emerging in the past few years has been cloud storage. Various companies are providing cloud storage services to their customers. Basically, with cloud storage you're uploading your data to servers of the company, which already has built a secure infrastructure for data backup. Some even offer these services for free such as Google, however the space under free accounts is relatively limited.

Cloud storage is currently a good alternative to onsite backup, however some of the other issues arise such as the importance and classification of data, trust, loss of data and others. Cloud storage for now, is just another place to store your data. You would never want it to be the ONLY place you store your data.

Lastly, the performance of cloud storage is severely limited by the end user internet connection. The type of transfer that suffers the most is the upload of files. Most cable ISP provides in USA tend to offer very low upload speeds. When you have to transfer several gigabytes of data to a cloud via network, it can take hours, days or even weeks until the content is uploaded. Also, because it is relying on a network connection, file corruption is also possible, as the downloads and uploads don't always complete properly.

Backup verification

One of the most important steps that are often missed and overlooked by most system administrators is the actual backup verification. You may have already setup your backup scheme, and you're constantly seeing new backups take place. However, did you make sure the backup actually works? Did you make sure the integrity of the backup is not in any way compromised? Many backup utilities of today have verification checks built in, however it's always a good idea to try and occasionally restore a backup, even on a healthy system. The purpose behind this is to test the backup process periodically and make sure no new bugs have come up to cause backup failure. Having countless gigabytes of backups is pointless if none of them can be restored properly.

Also, very important to backup healthy systems. Do not backup infected systems, corrupted files or systems that have malware, spyware and other infections. If you restore these backups, you will essentially restore all the infections. You want to make sure systems are healthy before any recovery takes place.

So where does one start with backup?

You don't have to invest thousands of dollars to backup your data. You can start with simple setup, which is just having 2 hard drives in your computer instead of one. Right there, you've decreased your chances of data loss. If you frequently copy file over to the other drive, you have eliminated a possibility of a single drive failure destroying your data. Of course, the more drives the better, the more storage controllers the better. If you have other computers data should be copied on them as well. If you have external hard drives or cloud storage accounts, you should also send your data there. It all depends how important your data is to you and how far will you go to protect it. You can even make tape backups and keep them in a vault if necessary.

It's important to plan ahead. Warranties, terms and conditions, replacement procedures and all those should be known ahead of time. You should be aware of how long a motherboard replacement may take, so that you can prepare for it adequately. Some motherboards take 2 weeks to replace. Others may take 6. You want to have proper redundancy in place so that you can resolve any issues ahead of time. If needed, it might be worth investing additional funds and purchasing spare parts to have on hand, should the downtime simply not be possible. Of course, you can never prepare for all possible scenarios. But it's a good practice to at least eliminate most common problems such as single part failures, data corruption, verification and downtime resolution. After that, there are still rare events that can cause downtime and data loss, but once again, chasing those down may require lot of funds and a lot of time. If you don't have that, try to do the best you can within your own means.