El Reg has teamed up with the Storage Networking Industry Association (SNIA) for a series of deep dive articles. Each month, the SNIA will deliver a comprehensive introduction to basic storage networking concepts. The first article explored data protection. This second one looks at building and operating a backup system. Part 2 …
This article is a good overview of options, with the BIG benefit of not focusing or trying to sell a particular system. Good show :)
With the Tape vs Disk argument, very balanced. I think, however, the "30 years" for tape is a bit generous, considering there have been studies showing that even under ideal conditions, tape degrades and leaves very little data (20% in one study) actually usable after even just 10 years. Combine that with requiring a tape drive (with interface!) that can read the tape. The same argument can be said about hard drives, but ask yourself how many computers still have IDE interfaces and then ask yourself how many computers have Ultra 160 SCSI interfaces (that still work). Not to mention you can get an external IDE USB enclosure whereas a SCSI tape drive equivalent is rare (or non-existent). Basically, the new ($2000+!) tape drives running SAS as an interface likely will still be connectible in 10 years, but I'd be more willing to bet on a SATA protocol. I would also be fairly certain a hard disk drive will last 10 years over a tape, if merely for the robustness and impermeability of the enclosure. For those thinking "flash drives would be awesome!" the answer is no, they wouldn't be. I believe the charge leakage of a flash cell is 5 years. Of course, the wear leveling algorithms of SSDs would move that cell data long before the charge diminishes to unreadability. Backups are meant to be cheap anyway, that's why tapes have lingered for so long. With 1.5TB disks dipping to the $80 mark though, and LTO5 tapes (1.5TB uncompressed, 3TB at the optimal 2:1 compression) at $70, it's almost a no-brainer to get the HDD, especially if your D2D software does compression too (which can get higher than a measly 2:1 depending).
As for types of backups, I personally am using a synthetic full with incremental forever method on D2D. For our archive off-site, we dump a full D2D (hence the need for synthetic fulls). Of course, the thing cautiously avoided in the article is the overwhelming cost of some of these backup methods (such as a tape library). Many small businesses could get away with a rudimentary tape backup system, or better yet, the good-ol' external HDD. Their DBs could be shut down at midnight for a full file-level backup (or sql dumped), and their file server copied. Perhaps an Exchange store exported. However, as far as dedup goes, only the file server would benefit from file-level dedup, and the exchange store and DB would highly benefit from block-level dedup. The more monolithic file stacks (think Sharepoint, SQL Server, VMs, etc) you have, the more block-level dedup can help with your on-disk storage.
I will say this: if the total size of your full backups is over 3TB, your benefit from tape backups significantly improves though :)
I'm not impressed with using terminology that will create confusion. Data Protection is a term that has so far has been used for protection personal information from disclosure - *NOT* a good idea to abuse those terms.
The term "data protection" has been used to apply to backup and archive (amongst other related technology) for a long time, nothing new here.
It shouldn't be surprising that the term has been used elsewhere, though, "data" is a rather generic term.
How is Silent Corruption handled? It is similar to ECC RAM errors. Every once in a while, your RAM bits get flipped by random (current spikes, cosmic radiation, etc) - that is why you use ECC RAM. Exactly the same problem occurs on disks too - bits flip on random. How is this Silent Corruption handled?
Recent research shows that no filesystem (except ZFS) handles Silent Corruption successfully. Nor do hardware raid. Here are research papers (including papers from CERN) on this:
Someone explain me how this bit is an advantage to push data across the wire:
"tapes can be physically removed and transported and this is a big advantage compared to disk"
The entire logistics bit is a nightmare and its less safe then a vlan/dedicated network.
Not to mention restore times.. Lets fire up the van, drive and get the tapes, can we even find the tapes? lets drive back and start restoring.
- Vid Hubble 'scope snaps 200,000-ton chunky crumble conundrum
- Updated + vids WHOA: Get a load of Asteroid DX110 JUST MISSING planet EARTH
- 10 years of Facebook Inside Facebook's engineering labs: Hardware heaven, HP hell – PICTURES
- Very fabric of space-time RIPPED apart in latest Hubble pic
- Massive new AIRSHIP to enter commercial service at British dirigible base