Your SQL Server setup is a mission-critical asset that you cannot afford to suffer an outage. In order to preserve uptime and plan for unforeseen catastrophes, you need a solid backup strategy in place.
There are a few things to consider when putting a backup plan for SQL Server together, so let's go over the main talking points and empower you with the information you need to deliver unshakable availability within your database infrastructure.
Backup types explained
First of all, it's worth discussing the fact that there are actually different types of backup and that this is very much not a one-size-fits-all solution to your downtime dilemmas.
There are three main types of backup to take into account:
As the name suggests, this backup type involves creating an exact replica of the information stored on your server so that it can be restored in its entirety if the original copy is compromised for any reason.
Full backups are all-encompassing and thus take time to complete. They also only give you a replica of the database when the backup was made; any changes that have happened in the interim will be lost if you need to implement recovery.
In general, full backups are better suited to smaller SQL Server deployments, or else they can become unwieldy.
Speedier than a full backup, a differential backup simply looks at what alterations have occurred within a database since the last copy was made and stores these.
Transaction log backups
Sometimes it is not enough to merely restore the data contained within a server after an outage. In any case, you want to roll things back to the state that existed before disaster struck, a transaction log backup will be required.
This lets you guarantee complete continuity when restoration occurs or when you switch from your main server to a secondary backup.
Leveraging SQL Server AlwaysOn Availability Groups will let you achieve this seamlessly within Microsoft's own ecosystem. Ultimately a combination of several backup types will probably give you the best balance for your own needs, which of course need to be taken into account during the strategy-building process.
Your plan for backing up your SQL database needs to factor in how and where server hardware is located.
If you are hosting it on-premises, then calamities like disruption to the power supplied to the building or things like fires, floods, and other natural disasters might punish you for putting all of your eggs in one basket.
This is why plenty of businesses choose to either collocate their hardware at data centers based in a different area or host their SQL deployment entirely in the cloud to alleviate themselves of the responsibility of keeping the infrastructure up and running.
It is best to hedge your bets and neither rely solely on an in-house setup nor a remotely hosted database configuration, both from a backup and continuity perspective, and when it comes to managing costs and streamlining monitoring and maintenance.
Scheduling implications & automation
We discussed briefly the idea that full backups might be unwieldy for larger databases, and this ties into the disruption that all types of contingency planning can bring with them.
Even lighter weight differential backups will put a strain on the database and impact service availability and performance as a result. This is where scheduling comes into play as part of your strategy. Your backups cannot be made in a way that compromises the day-to-day operations of the server, so you need to look at the analytics and schedule them accordingly.
Identifying the points of the day or week during which SQL Server activity is at a low ebb is the easiest way to work out when backup processes can be put into action, whatever form they take.
There is a balance to be struck between database disruption and the accuracy of any backup you make. This is why it can be helpful to stagger your backups and use several of the types mentioned earlier.
For example, you might carry out a full backup once a week during an off-peak period, then implement differential backups after each working day so that you have all your bases covered while still preserving server performance.
Aspects like the size of your server and your available resources should also influence your scheduling decisions. And if you find that the schedule you initially decide upon isn't working, don't be afraid to change it.
The importance of testing & verification
Talking of scrutinizing your fledgling backup strategy, it's also vital that you rigorously test the solutions you put in place and verify that the data being stored is comprehensive, accurate, and uncompromised.
If you don't do this, you could find that it isn't in the ideal condition when you need to restore data after a loss event or an outage. This is particularly pertinent when it comes to high availability applications of SQL Server.
Planning to switch from a primary database to a secondary backup the moment that unplanned downtime occurs is all well and good, but not doing your due diligence by testing this part of the infrastructure is unforgivable.
Additional maintenance considerations
No SQL Server backup strategy is complete without proper consideration being given to the maintenance issues that can lead to downtime if overlooked.
Component failure is difficult to predict at a hardware level, but monitoring performance and looking for erratic behavior from your CPU, RAM, or I/O devices can let you spot the early warning signs that a full-blown breakdown is on the horizon.
From a software perspective, pinpointing troublesome processes in the OS and poorly optimized queries and fragmented indexes in the database itself is better than letting them run riot and potentially lead to the whole database grinding to a halt.
Tips for SQL Server maintenance
The most important piece of advice for any database administrator is that prevention is better than cure when it comes to maintenance. Also, monitoring server performance to troubleshoot outage-causing issues early is better than scrabbling to restore lost data further down the line.
Aside from tackling the aforementioned hardware flaws and complications with queries and indexing, one often-overlooked aspect of maintenance is the need to be aware of how server usage will change over time.
Uptime can only be maintained if the server is fit for purpose, so if you are pushing components to their limits and running out of space with no plan to upgrade, you are taking unnecessary risks.
Keeping others in the loop
Thorough documentation of any SQL Server backup strategy is a responsibility that has to be fulfilled comprehensively.
It is no good putting all sorts of plans and contingencies in place, only for them to be rendered useless if someone else in the organization has to work out how to implement them should you be unavailable.
Likewise, you need to keep everyone up to speed with how long it will take to restore full functionality should an outage strike. Employees and customers will be far more tolerant of downtime if they know it will be brought to an end within a specific timeframe, rather than being left in the dark.
Ultimately your backup strategy must be tailor-made for your SQL Server resources and your broader business aims.
Take availability tolerances into account and do not cut corners if uptime has to be preserved at all costs; otherwise, recovering from a database outage will be monumentally expensive, not to mention poisonous for your brand's reputation.