cecs - Fotolia
The seventh generation Linear Tape-Open (LTO) specification offers more than a doubling of the maximum compressed capacity to 15 TB and an increase in the data transfer rate to 750 megabytes per second.
Newly released LTO-7 also improves the bit error rate to a significant degree,
Shawn Brume, a business line executive for data protection and retention at IBM, and Laura Loredo, a worldwide product manager for HP's enterprise servers, storage and networking, recently spoke with TechTarget about the significance of the new LTO-7 specification and adoption trends with the Linear Tape File System (LTFS).
They claim the open-standard LTO Ultrium format commands about 85% of the tape market over proprietary formats from vendors such as
Is there anything special about the LTO-7 update other than the boost in the capacity and data transfer rate?
Brume: The bit error rate has improved over previous generations, and that can actually be significant. To the common customer, it may not look like there's been a big difference, but the bit error rate drives reliability and availability of data. … Cloud providers say they have nine nines of availability. ... By moving to the higher bit error rate on tape, we actually move tape into the availability rate of 12 nines on a single copy of data when you calculate out availability based on bit error rate.
So, it is a significant jump in the long-term reliability of data to be written correctly and be able to be read back correctly. And since tape is the lowest cost target for data -- that often means data is going to stay there forever -- having that availability and that reliability is very important.
To what degree did the error correction code improve in LTO-7 compared to LTO-6?
Brume: The correction codes change every version. The bit error rate in LTO-6 was [10Ex17] and now in LTO-7, it's [10Ex19]. It would be 1 with 17 zeroes after it versus 1 with 19 zeroes after it, as far as the number of disks written without error. So the method and the capabilities of the error corrections during write and read operations has been significantly improved enough based on coding and the manner in which the algorithms are run to make that statistical claim of availability, or should we say, reliability of data.
They're both large numbers. Is the difference significant?
Brume: They both are significantly large numbers, but the difference means the difference between writing one petabyte (PB) of data without an error and writing 100 PBs of data without an error. And if you compare that to disk, you're looking at hundreds of terabytes (TBs) vs. hundreds of PBs.
For what types of customers will the bit error rate improvements be meaningful?
Brume: High-end customers have always written into the 10s and 100s of PBs of data: financial industry, oil and gas, those collecting geospatial data. They've always written petabytes. But what we're seeing is new use cases where they're writing not only PBs but exabytes and zettabytes of data, which includes cloud.
When you look at cloud providers that are offering storage, those CSPs [cloud service providers] need a reliable medium. … Google Nearline and Amazon Glacier say this is a place to put your data. It will survive forever. Well, those CSPs need the lowest cost storage in order to turn margins. What better way than to put it on tape and know that your reliability is three orders of magnitude better than what you can do with disk.
Other use cases that we're seeing growing: media and entertainment [M&E]. There are rumors that both Star Wars 7 and the next Avatar are going to take over a PB of data to do all the transforms and everything in order to release those to market. That's a lot of data in the M&E industry.
How is in-house tape faring vs. cloud archiving?
Brume: Cloud archiving's new. It's making its inroads. They still don't have nearly as much data within cloud as there are on-premise installations, or what we might call on-premise clouds, for some of these other vendors. But some of the majors within the industry have already talked that in their storage infrastructures, they are using tape. … The fact is, tape is more alive even in the latest use cases -- cloud -- than it's ever been before.
What kind of traction are you seeing for LTFS?
Brume: We know that it took off extremely well on its introduction into the media and entertainment industry. That was mainly because with media and entertainment, they're very much early adopters. But what's happened since then is those infrastructures have grown. The need for open, non-proprietary standards of the data on tape is really pushing customers to say, "Hey, I need a format that's easy to use. How can tape help me?" And LTFS has been the great savior there in that customers look at it and go, "OK. So what you're telling me is I can view my data like it is a file system interface. I can drag and drop. I can do adds and deletes and everything I need to do, and when I tuck that data away, I don't have to keep the application that wrote the data." That right there is a huge savings because they don't have to keep this huge database.
Loredo: It has grown very big in media and entertainment, but it's going to all the different verticals where they have a need for long-term retention of data, and they have big amounts of data. So anything that requires video, like video surveillance, medical images, media and entertainment, they're creating huge amounts of data that they need to retain for very long periods of time. It's ideal for these verticals.
Brume: Does it mean it's their global standard? No, but I think a great reference on how well it's expanded is in late 2013, early 2014; Discovery Channel made the mandate that if you were going to send them video, you must send it on an LTO tape formatted in LTFS. That's a pretty big statement considering that they're one of the largest ingest houses for raw data.
Have you seen any change in the last year or so in the way that people use tape? Are they using tape as a tier with flash- and disk-based storage arrays?
Brume: It certainly is used with the flash and disk systems, and there's a lot of information out there where you can see people are recognizing the value even of a flape solution -- flash and tape -- because tape is going to give them that lowest cost storage for that 80% of data that rarely, if ever, gets touched that they have to keep.
What we're actually seeing is a lot of people are coming back and saying, 'I have to store that data, and I have to store it forever, and as I move my infrastructure, because now we're starting to see an economy turnaround, I need to figure out how to implement tape better in LTFS.'
What are your predictions going forward for tape?
Brume: You're going to see tape really stabilizing. You're going to see the industry of cold storage growing, more cloud providers recognizing the need, more social media recognizing the need for tape infrastructures. I mean, 300 hours of video uploaded to YouTube every minute? That data's got to go somewhere, and ultimately, if they're going to keep it, most of it's not going to be accessed.
Are you going to see giant revenue growth? No, it's not about giant revenue growth. If you look at the capacities of the tape and the ability to deliver that, what you'll see is a flattening of revenue, but you're going to continue to see the exabytes of data shipped every year continue to grow significantly. So, you'll see probably less customers at the extreme low end using tape and at the extreme high end [they will be] adopting even more tape.
Tape remains important for archiving
Explore storage possibilities for deep archives
Energy efficiency of tape vs. disk
Estimating how long backup tape can last