Object storage systems are typically software-based solutions that run on commodity servers, which are clustered to create a single pool of storage. As a result, object storage solutions tend to be very cost-effective from a-price-per-GB perspective. These systems also have excellent data protection and preservation capabilities. But are they ideal for backup storage?
The Challenge with Using Object Storage for Backup
A storage system needs several critical capabilities for it to be a viable solution for data protection. First, it has to be very cost effective. Second, it has to be scalable, able to store potentially petabytes of protected data. Third, it has to be fast – not as fast as a primary storage system, but fast enough to be able to ingest data rapidly. A new requirement, thanks to features like in-place recovery (where the backup software mounts the recovered volume on the backup hardware) is a different type of performance. Modern data protection storage has to deliver excellent random IO.
These last two requirements – rapid ingest and suitable random IO performance – may be the Achilles’ heel of most object storage systems, but that does not mean these systems don’t have a role to play in the backup process.
Any data protection storage system that meets the rapid ingest and random IO performance demands of the environment is going to incur costs associated with delivering that performance. But the actual size of data that make these extra performance demands is relatively small. The performance is only needed as new or modified information is received by the data protection storage, or when an application’s volume is instantiated on it. All the other data on protection storage, well over 75% of it, does not need any measurable performance. It is just going to sit there and be stored.
The solution is to leverage both. Invest in a well-performing protected data storage solution for rapid restores and an object storage system for long term retention. In this design, IT will store the most recent few months’ worth of backups on protection storage and then archive older backup jobs to the object storage system. In this way, IT leverages both systems for what they do best while at the same time keeping costs under control.
An additional value of object storage is that it can do far more than serve as a repository for old data from protection storage. It can also store old data from primary storage, replace the use of NAS for home directories and user data storage and, of course, be used by modern applications designed to natively leverage the S3 protocol common in most object storage solutions.
Object storage should be viewed as a complementary technology to data protection storage, one that can limit the growth and lower the cost to purchase, operate and protect both primary and secondary storage while also laying the groundwork for modern applications. The savings allows IT to invest in higher-performing primary and secondary tiers while capacity expansion is primarily limited to the cost-effective third tier.
To learn more about using object storage to compliment protection storage, check out our on-demand webinar, “3 Steps to Stop Data Protection Sprawl”.