The Unofficial Backup Admin's Handbook Part 4

Storage Mediums

So let's now cover the various different storage mediums out there and some of their attributes:

Disk (non-deduplicated) - SAN or Local disks
Hard disk storage (either spinning or SSD) available to a backup server as storage space. Images are stored in their original size from the backup client.

Best used for short term storage (such as staging) with minimum CPU and memory overhead

Costly for longer term storage

Very fast backup and restore (as data can be stored in a nonlinear format)

Can be replicated offsite for Disaster Recovery/Business Continuity purposes

Higher risk of data loss due to server tampering or virus/malware corruption (varies by platform)

Disk (deduplicated) - SAN or Local disks
Hard disk storage (either spinning or SSD) available to a backup server as storage space. Images are broke up into blocks, "fingerprinted" to verify unique data, and only unique data is stored. All duplicate data is reduced to a single copy and references to that copy are made in the image database.

Best used for short to medium term storage (such as an onsite copy, in my opinion <60 days)

Increased CPU and Memory usage on storage server or clients for deduplication and fingerprinting processes as well as maintaining the database. The deduplication database is generally recommended to reside on SSD storage, while the main data files can remain on spinning disks

Costly for longer term storage, more value than non-deduplication disks due to increased data density

Comparatively fast backup and restore (as data can be stored in a nonlinear format)

Can be replicated offsite for Disaster Recovery/Business Continuity purposes

Higher risk of data loss due to server tampering or virus/malware corruption (varies by platform)

Replication jobs between two deduplicated disks tend to only transfer unique data, reducing the replication window time and bandwidth required

Tape Media
Tapes used for backup storage. Capacity depends on the format used, and often must correspond to a particular tape drive, with some backward compatibility supported (LTO drives can generally write one generation back, and read up to two generations back)

Generally the slowest medium for backups and restores

Data is stored in a linear format, so to access non-sequential data requires scanning through the tape

Arguably the best (and cheapest) method for storage of data long term as tapes are not required to remain in the library, consuming space and power in your system when not needed

Biggest drawback is risks due to strong magnetic fields and format changes

Data can be stored either deduplicated or non-deduplicated (commonly referred to as "re-hydrated")

Large sequential backup jobs (such as large database backups) are also good candidates for storage to tape as they tend to not require much scanning through the tape

Tapes can be sent offsite to either a Disaster Recovery location or a third party facility (i.e. IronMountain) for storage, reducing risk of loss due to environmental events

Lower risk of data loss due to virus/malware/tampering due to data not being online when not in use

Optical Disk Media
Does anyone still use this? This would be rewrite-able media such as CD-RW, DVD+/-RW formats which can be erased and reused

Attributes are similar to non-deduplicated disks, although capacity is extremely limited

Very delicate media compared to others - very scratch prone, although not susceptible to magnetic fields.

Cloud
You knew it had to be mentioned. Storage of data in either an on-prem or offsite cloud provider using various cloud APIs (such as Azure or S3)

Attributes are generally the same ask Disk Media, although data can be deduplicated, and can have dates assigned at the cloud storage level to allow for required retention (to prevent either accidental or intentional erasure from the backup application)

Cloud providers have multiple tiers of storage ranging from the very fast EBS/S3 storage media on disks, to something similar to Glacier, which is comparable to tape media on a cloud provider

Offsite cloud storage requires the bandwidth to send off and recall data back from the vendor, so speeds will vary

WORM Media
Media with WORM attribute does not allow erasure or data to be overwritten. WORM stands for Write Once Read Many. Other attributes match the above media details

Most commonly used to legal holds or any data backup/archives that require protections against accidental erasure/overwrite, even if tapes are moved from one product/environment to another

Tape (WORM)
Tape medium with various form factors (LTO is most common type now) Density/Capacity depends on the format of tape (LTO3, LTO4, LTO5, etc, 

Optical Disk (WORM)
Examples of Optical WORM media would be CD-R, DVD-R, DVD+R.


This list is not all inclusive but should cover a majority of the media encountered in day to day operations.


Main Index
Part 5: Destinations

Comments