In one sense, cold storage has been around for decades. It’s called “tape.” Because what is cheaper than keeping decades’ worth of storage in a controlled-climate vault? No spinning media, no energy usage, no retention concerns unless Iron Mountain suddenly goes out of business. And if your only cold storage requirement is that it be cheap and that you can prove the tape exists, then you’re golden.
Except when you actually need to restore something quickly, like a file share-full of irreplaceable marketing images. Or a year’s worth of digitized sales contracts. Or – and this is really sobering – a 12-month range of email threads from 3 years ago for a big pending lawsuit.
Luckily the data is on tape. Realistically you need to comb backup catalogs to find the tapes to recover, issue the order to the vaulting company, wait on shipping, sign in the tapes, look around for an older version of the backup application, and find or buy a tape drive that works with the older cassette.
This is where modernized cold storage enters the picture. Older data has regained value for different use cases such as litigation, compliance, digital asset storage, even business analytics. Cold storage has changed in response to the need: inexpensive scale-out storage offering online data retention with RTO from minutes to less than a day.
Requirements for Updated Cold Storage:
- Inexpensive scale-out. Scale-out storage systems for production environments have big price tags because of high performance and redundancy requirements. Cold storage providers can build scale-out systems on cheap SATA disk and commodity components.
- Object architecture. Object-based storage also enables vendors to offer scale-out architecture at low prices. Unlike block storage SANs that must be attached to an OS, object storage is independent of the operating system. Data can enter the system via APIs or an online connection. Any file type can be stored in an object format, which makes it particularly useful for large unstructured data like archives, user-generated data, and pictures and videos.
- RTO tailored to use cases. RPO is not an issue in cold storage but RTO is. Waiting days (or longer) for tape restores used to be acceptable, but this is rarely the case anymore. Cold storage has to be responsive enough so that you can restore data within timeframes of less than a day down to a few minutes.
- Online/nearline access. Offline storage such as vaulted tape requires time to access the data. Online and nearline access keeps data immediately identifiable and accessible to users. Recovery time still figures into the complete RTO, but there is no waiting on the physical storage.
- Cloud integration. The cloud is not a requirement for cold storage, which some companies build on-premise for recovery or security requirements. But cloud providers like Amazon and Google offer cold storage services, while many MSPs offer cold storage-as-a-service using public or private clouds. And even on-premise clouds may use cold storage in the cloud as another tier.
Use Cases for Cold Storage:
- E discovery and compliance audits. These are two of the biggest drivers for highly recoverable cold storage. It is challenging to retrieve large-scale eDiscovery data from tape, especially since matters often deal with data that is over 2 years old. Searching and recovering data from online/nearline cold storage can speed up ESI requests by magnitudes.
- Digital assets. The broadcast industry remains a heavy user of online tape libraries to preserve access to film and images. On-premise disk-based cold storage offers them a nearline storage tier between production storage and tape libraries. Another example of digital assets in cold storage is medical imaging, which produces massive data sets and whose availability is highly regulated.
- Video surveillance. This large and growing industry retains video images for years for security and litigation reasons. Cold storage allows vendors to search and recover dated videos while saving money on scale-out storage systems.
Tape is by no means dead; tape libraries are an important component in cost-effective computing such as active archiving. But for cold storage where inactive files need to remain accessible, look to object-based, scale-out and cost-effective cold storage.