Oracle uncloaks 5TB tape Oracle is increasing the capacity of its T10000 tape fivefold, doubling its I/O speed, and enabling a 5-exabyte tape library. The T10000 C (T10K C) tape has a 5TB native capacity – the highest in the industry – and the associated drive has a 240MB/sec throughput. In comparison, the current T10K B format, introduced in August …
Lets see. Current price that I see for T10000 @ 500GB capacity is about $137.95 each. I can get a 1TByte hard drive for under $100. If I have to restore from this, I don't have to 'wade' through the tape.
If they don't get their price at a useful price-point, forget it. I'll just use a RAID assembly with removable drives and archive the hard drives.
HDDs are not well regarded for archival purposes in the seismic processing business. If I want to see my data 15 or 20 years from now I throw it onto tape. The cost of the tape is immaterial compared to the cost of the research, and historically tape has offered me better stability and fewer points of failure.
Good luck finding a tape drive that will be able to read your tapes in 15-20 years time! Unless you maintain a legacy tape drive environment for all that time your tapes will be next to useless.
You could always keep migrating the data off them on to newer tape technology I suppose.
It's for the big boys. Once you get to 1Pb+, having to manage all those spinning discs, suddenly makes tape the more logical choice for long retention. Factor in power and cooling, the bete noir of the DC Manager and Green Crowd, and it is actually HDD that starts to look out of place. Add back decreasing prices of flash memory for better data availaibility, and HDD's role could be relegated to that of interim cache between archive and flash.
As someone else mentioned, managing hundreds of removable sata drives would be a nightmare for policy and practicality.
Have decided that backing up to tape is so last year... Is busy putting together a Linux server to house our archives, and do a backup to tape once a month to save on tape drive wear and tear.
It is a schlepp though - but if hard drives start to get cheaper than tapes, then surely why not archive it to hard drive instead of tape? Quicker response time... but the drawback of hard drives is growing defects.
So tape, as long storage medium, still have its advantages.
Can still remember the dinky 40Mb tape drives you used to connect to the floppy drive controller back in the days.
What we need is a HDD that will never fail or lose its data - that can keep the data safe for 20 years. By that time you won't be needing that data anyway. But wait... hard drive interfaces do change, by that time you won't be able to read from the hard drive(s) at all - look at the older type of HDD's once found in IBM XT's...
Same can be said of tape - by the time you want to read from tape, no suitable tape streamer can be found (or one that's still in working condition...)
And in the meantime the amount of data you'll need to backup doubles in size. Subset to Moore's law anybody?
Makes one want to ESC the rat (backup) race...
Just doesn't work for any serious length of time. Harddrives will break and you won't have the right interface to connect to them. Tape drives will break down or the interfaces won't be available or the software you used doesn't exist any more.
The only way to archive data is optically. Those accounting files you might need in 40 years? You need to photograph them and store them on film. That way they can be read by anyone with a lamp and a magnifying glass.
SW that uses proprietary data format to write (Symantec springs to mind) is biggest concern for migrating data. An open format such as tar makes sense as it can be read anywhere.
As for reading a 15 year old tape, this still happens in the MF market and with some savvy operators in the OS sector. Of course with an open format it is quite easy to migrate through tape generations. People on LTO 1,2,3 have migrated to LTO 4, 5 with little operational impact, so that's 12 year straight away. Cost wise a few tape drive upgrades in a library, or even a new library are still far cheaper than cost of powering and managing that same data on an expensive HDS, EMC array for that period of time.
BMW Gmbh last year finally decomissioned a tape silo and migrated data to newer formats, after 18 years non stop operation.
Show me a 99.9999% array that is still in production at the data centre after 5 years. Show me a Data Domain appliance that has been running anything approaching these sorts of duty cycles.
This is sort of reliability that tape offers and cannot be matched by HDD:
http://www.enterprisestrategygroup.com/2010/12/nersc-proving-tape-as-cost-effective-and-reliable-primary-data-storage/
Until there is a quantum change in storage technology tape has a role, and that is growing in current market.
It takes me about 6 minutes to restore a single file from LTO5 tape and most of that time is fetching the tape from the data safe/putting it in the changer.
Decent backup software knows where on the tape the file you want is and will skip straight to it.
It doesn't have to be pricey - www.bacula.org is getting a LOT of converts from expensive packages because of its usability and flexibility and it handles disk-based backups as easily as tapes.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
