Cloud backup is a powerful tool for organizations looking to safely and securely store data, both for quick and easy accessibility and for long-term archiving and to meet regulatory compliance needs. But Cloud backup requires a little more insight and attention than just simply offloading data into a repository.
As Rackspace’s Chuck Thier points out, you have to know what you need first, then assess why you need it, and then follow a few simple best practices to make sure you’re getting the most out of your backup.
Thier points out that it’s important to understand the difference between throughput and random I/O so you know what type of block storage is best for your organization. “While throughput is very important for writing sequential data to your drives — logging, streaming data or basic file access – throughput is roughly the same for both SATA and SSD drives,” he said. If that’s all you need, then sticking with a standard volume SATA solution is the way to go.
Random I/O is a different story, he said. For applications like databases and NoSQL servers that more quickly write data to random parts of a disk, speed is essential. In this case, an SSD offering will make much more sense, Thier said.
First and foremost, you should know when to implement RAID, and when not to, he said. If you are in a High Availability (HA) environment and your data must be available at any time, then setting up a RAID array as you would with a dedicated environment makes sense. Says Thier: “Usually in a mirrored configuration, to help provide extra redundancy and durability. This means that if a drive goes away or fails, you can easily replicate your data and replace another drive in the RAID array.”
This may seem obvious, but you also must back up your data, he advises. Whether you are using RAID or not, it is always a good idea to back up your data. You also can use Cloud Backup to make a file-level backup of your entire block device for recovery purposes in the event of a disaster.
Once you’ve determined which backup option is right for you, keep in mind these ten best practices put together by CA.
1. Think of local protection as the first line of defense. The public cloud is great; it offers virtually unlimited server and storage resources and can be extremely flexible, allowing you to add and remove resources as needed and use operational budgets (OPEX) instead of capital budgets(CAPEX). But don’t go getting rid of your on-site solution just yet, especially if you need speed and performance. Most of the time, the best performance will be delivered by using resources local (on-premise) to the systems and data being protected.
2. Identify the systems and the dependencies of those systems that are critical to the business. If local protection is your first line of defense, then public-cloud-based protection would be the second line of defense. You should prioritize the servers and data that need offsite protection by how critical those systems are to your day-to-day operations. Also, make sure you understand the dependencies of those critical services, such as which databases and middleware are connected, and make sure they are also protected in the public cloud.
3. Don’t think only of traditional backup for disasters. We always consider the possibility of large-scale disasters and of losing entire servers, but many recovery operations are required after a disaster happens. In addition to standard recovery, make sure that you can perform snapshot or point-in-time image recovery to ensure you’re getting ALL your data back.
Also consider the ability to use replication technologies to provide continuous data protection locally and in the cloud, for those mission-critical systems discussed earlier. Even high-availability software solutions can be used with a public cloud for automated and push-button failover for the most critical systems and applications.
4. Think about how you want to restore data and back up to meet that goal. Backing up the system and all the storage will protect everything on that OS instance, which is perfect for when you need to restore the entire environment using bare metal recovery scenarios. If you are restoring a service or database — like Microsoft Exchange, for instance, or even just a single e-mail — you will have different backup requirements. So think about what you might want to restore, then make sure you are backing up in a manner consistent with that goal.
5. Backup at the hypervisor level may not always be enough. Virtualization has introduced many awesome capabilities, including the ability to perform backups at the hypervisor level of the virtual machines(VMs). But this type of backup typically will limit your restore to a VM-only level or, at best, to files within the VM. For rich services within the virtual machine, you should run backup agents within the operating system instead of just on the virtual machine for a high-availability solution.
6. Use the cloud for long-term backups. Data is stored for many reasons, the most important of which is long-term archiving for corporate compliance needs and to meet regulatory requirements. Maintaining long-term disk-based backups on a company’s resources can be very costly; maintaining long-term backups or archiving data in the cloud can be a much more cost-effective solution.
7. Ensuring security of the public cloud data. Securing your organization’s data is a major consideration. When choosing a cloud backup provider, you should always verify the security used in the solution—for example, the physical security of the public cloud locations, encryption of data at rest on the storage, and logical separation of your organization’s data from other organizations using the same public cloud backup provider. You should also understand how your data is sent over the network to the public cloud and what encryption is used to protect the data during transmission.
8. Running the recovery directly in the cloud. With your systems, applications, and data protected and with backups stored in the public cloud, you have protection if you lose an entire site. But where do you restore your systems if your site does go down? Consider running your systems in virtual environments in public cloud virtual machine hosting solutions using the systems and data backed up in the public cloud. This approach allows your operations to be up and running again even without your own datacenter.
9. Unified backup and management. Most organizations that leverage a public cloud for backup will still have a local solution. This means you’ll have a different management system for each type of solution; instead, consider one, unified management system via a hybrid model that will enable a single management approach.
10. Test the processes periodically and any time a significant change occurs in your infrastructure. The best solutions in the world will fail if you don’t know how to use them correctly—and if you don’t perform regular tests to ensure restore processes work and the data protected is valid. Get into the habit of performing regular tests.
Cloud backup involves more than just offloading data into a separate repository, but once the initial workload is handled, cloud backup can provide a secure, cost-effective, and efficient way to maintain your data stores now and for the future.
Sharon Florentine is a freelance writer who covers everything from holistic veterinary care to data center technology and occasionally blogs for cloud provider Rackspace Hosting.