Deriving Backup Solutions from Customized Needs
There are plenty of articles that focus on the business benefits of the new technology, so I won’t re-hash that subject. Instead, I will share my thoughts on making the technology change and what is involved. There are many aspects of implementation that need to be considered and planned prior to adopting the new generation of backup technologies.
A One-stop Enterprise Scheduler
First and foremost, make sure you consider all of the automation tools involved in your enterprise backup solution today. Most IT shops depend on a suite of tools that work together to make backups happen. However, many of the new technology backup solutions bring functionality that make it tempting to get rid of these traditional tools in favor of simplifying the enterprise landscape. If your enterprise has been around for a while, it most likely utilizes a job scheduler to orchestrate business process activity with backup activity. For example, IT operations have to perform database backups which require coordination with month end accounting cycles. Running out-of-phase with the business process results in useless backups that cannot be relied upon to restore data to a point in time related to closing the books. Be careful, don’t make the assumption you can replace the enterprise scheduler with a scheduler included in your new backup solution. Instead, review the functional requirements of your existing scheduler and confirm the backup scheduler can perform all of the same functions. Enterprise schedulers are commonly used to execute business processes via scripts or API’s on application platforms as well as kickoff backups, most backup solution schedulers can’t do all of these things. You may end up with multiple schedulers in the landscape and have to work to integrate them to work together, which can create unnecessary complexity and confusion for your IT Operations team. Consider using a single enterprise scheduler that can orchestrate all jobs, including backup jobs. Investigate the availability of an interface that can work between your enterprise scheduler and the backup solution.
Identifying use cases helps to understand the detailed requirements for each type of backup
Data replication is the replacement for offsite tape storage. By having an electronic replica of backups at an alternate site, the threat of losing the only copy is mitigated. Think through all of the data storage locations that will house your enterprise backups and make certain all are replicated to an alternate site. To make replication feasible most solutions rely on data de-deduplication to minimize the network bandwidth needed to replicate data. There are different kinds of data de-duplication and each has benefits in different situations. Consider how the full lifecycle of the backup data may play out before you implement data de-duplication. Data de-duplication can be carried-out on the client (host) or it can be accomplished off-line which happens after the backup process is complete. Keep in mind Data Center backup and restoration timing and calculate the restore time incurred when using real-time de-duplication. Conversely, leveraging de-duplication prior to replication over a WAN link may be the key to having a cost effective solution in terms of WAN bandwidth consumption. Get it wrong and you will impact the business with an unacceptable restore delivery time.
Next, make sure you analyze your enterprise backup needs and create a finite list of backup use cases. Identifying use cases helps to understand the detailed requirements for each type of backup, which in turn helps your team design backup solution patterns that will streamline how backups are done. For example, backing up a Virtual Machine host image can be done inside the hypervisor layer, but doesn’t always meet the business needs. Sometimes VM’s rely on non-virtualized disk volumes, which are not included in the image backup. This type of analysis should be used for all of the use cases and should result in a short list of backup scenarios and design patterns that support the requirements. In short, don’t try to use the same design pattern for all backup scenarios, sometimes a new design is needed to avoid gaps or performance bottlenecks. A short list should include VM’S with Virtual disk volumes, VM’s with direct disk volumes, Operating Systems, and Databases. Also consider where multiple flavors of these technologies may require different approaches. For example, backing up MS SQL may be different than Oracle RDBMS in terms of how the technology works.
Enterprise Backup of remote site servers brings another level of consideration to adopting this technology. Don’t overlook the details on how to perform a restore of a remote server. When the de-duplicated data resides at different location than the server, the restore will likely be very slow over a WAN link. Can the business tolerate a 48-hour restore process? Yes, the backup process works well over the WAN, and makes financial sense, but backups are done so that data can be restored and the business can continue to run their systems. There are many options on how to satisfy the business requirements for this scenario, including planning to build a replacement server at the replication site and shipping it to the production location in the event of a disaster. Another approach is to have a local copy of the de-duped data onsite, so a local resource can restore directly to the server without traversing a WAN. Pros and cons exist across the board, so be sure to think through the business requirements and then design a standard solution for each unique use case.
Adopting the latest in enterprise backup technology is absolutely the right thing to do. Implementing this technology requires careful review of how your enterprise backups operate in the context of business processing. Build a solid foundation that considers the integration of job scheduling with backup scheduling. With this foundation in-place, think through all of the backup scenarios in terms of common use cases and design backup patterns to support them. Leveraging a short-list of standard design patterns will increase quality and performance, which are key drivers for adopting this technology.