Spectra Logic recently hosted a virtual presentation on the history and future of tape. During the webinar we took a look at
tape library innovation and how organizations have harnessed tape technology to achieve dense, cost-effective and reliable data storage. The following questions and answers recap highlights covered in the recent webinar.
LTO-8 will only write/read on current and one previous generation, whereas LTO-7, LTO-6 and LTO-5 read/wrote two generations back. Will this trend continue?
We expect the “single previous-generation” read/write compatibility to continue moving forward.
Is the 10
19 reliability of LTO-8 vs. 10 21 of IBM’s TS11XX an academic exercise… isn’t LTO as reliable as needed, short of malicious acts by bad actors?
Many would say, “Yes.” The bit error rate on both LTO and TS Tape Technology is outstanding, and several orders of magnitude better than capacity disk drives. When combined with making multiple copies of data on multiple tapes, which is very reasonable given the low cost per gigabyte of the tape media itself, it is virtually inconceivable that bit errors will cause data loss using either technology.
Is it concerning to Spectra Logic that there is only one tape drive manufactured now? What if IBM stops making tape?
It is true that there is only one manufacturer of tape, IBM, who makes both the TS Tape Technology and LTO
Tape Technology. There are several reasons that this is not of concern to Spectra. Tape is still a profitable market, and shipments have increased year over year from 2019 to 2020. If IBM decided to stop manufacturing tape, the tape division would be sold. As much as the business model for tape itself, tape is a very strategic product. It is used (and demanded) by the largest data centers in the world. These centers also use computing, networking, other forms of storage, software, etc. Providers of tape (those who manufacture it as well as those who integrate it) see that it fills a very unique and important role in the industry and would prefer their customers not have to go to another vendor to acquire that part of their data archiving strategy.
Tape Users in the Real World
Do places such as
Lawrence Livermore and NASA, with their huge tape libraries, have TWO such libraries, geographically separated?
While we don’t comment on specific configurations at any given HPC data center, we do see some very large data centers that have multiple site libraries. This tends to be the exception rather than the rule due to the multi-petabyte environments they operate in. It’s far more common to create the tapes at one site and send those tapes offsite for disaster recovery rather than have a large library(s) at a second site. The beauty of removable media, right?
I recall years ago that Google had to recover some Gmail customer data from backup tapes. Can you speak more to the big public cloud firms actively using tapes for large
Each large cloud provider operates differently, and they are usually very secretive about how their data centers are configured. We can say that we’ve seen multiple cloud providers who use tape for either cloud storage or cloud backup.
Tape and Industry Futures
Can you comment on how much time is typical between the fantastic capacity demonstrations in IBM’s R&D labs and when those capacities hit the mainstream market?
The most recent lab demonstration between IBM and Fujifilm showed 580TB being written to a single tape. The technology they use in these lab tests is used in current and near-future tape drives. The published LTO roadmap currently extends through LTO-12 and projects up to 144TB (native) per tape. As we tend to see two to three years between LTO releases, it will still be quite some time before we see a 500+TB tape commercially available. The real significance of these laboratory tests is to prove it can be done, explore the technology required, and fuel the ongoing roadmap for tape.
What is the future of data storage beyond tape? When will we begin to see those technologies on the market?
In terms of storage appropriate for storing large amounts of data for long periods of time – think archive – many storage technologies are currently under investigation. The more novel involve holograms, 3-D optical storage, 5-D glass storage and even DNA as storage devices/mediums. While these technologies are interesting, we don’t see them becoming viable, commercial storage mediums any time soon. Manufacturing ability, reliability and cost are all concerns which must be addressed to make them viable for the commercial market. Tape is the lowest cost, most reliable, most easily transported medium we have available today for archive storage and it will most likely retain that distinction for the future.
“The History and Future of Tape” webinar is a part of
SpectraLIVE, Spectra’s virtual conference program. Watch it on demand here.