Around the Storage Block
Showing results for 
Search instead for 
Did you mean: 

tNAS rule #1: know your workflow


MF.jpgBy Mark Fleischhauer, HP StoreEver Tape Storage Solutions Engineering Manager


When tNAS is the right answer to your archive data storage needs

I get this question from time to time: “Do you have anything that can help me store my files and content that I can’t afford to keep on my disk array?” My answer is, “Yes we do!” We have a solution we call tNAS, short for Tape-as-NAS. After explaining tNAS and its benefits, I then get asked, “How do I size it to meet my archive demands?” That’s when the workflow discussion comes into play. Here I’ll explain how the workflow impacts the sizing of the disk cache, the number of tape drives and the tape library capacity for storing archive data in a tNAS solution.


Quick tNAS overview

In its simplest definition, tNAS is just as you might expect. It’s tape storage that is accessed like a NAS mount point. As such, reading and writing data to and from tape is as simple as reading and writing data to a disk NAS. This makes it easy to integrate into your existing workflow; no special administrative request needed to access your data. The tNAS solution uses a host computer and some spinning disk (or flash) that present a NAS mount point. The disk “hides” the sequential access requirements required for tape for both reads and writes. Sizing this disk space and the tape library is dependent on the specific workflow and access needs required. Let’s take a quick look.


Identifying the kind of workflow

For simplification purposes, let’s define two different types of workflows:

  • Project-based workflow: This type of workflow is one that is based on a user sending a collection of data and/or content at one time to the tape archive. Data dumps occur when a project is completed and needs to be saved/archived to free up space for the next project. For example, a movie project or TV show with all the intermediate files and metadata files may be archived once the raw footage is combined and edited to create a final master copy.
  • Steady-stream workflow: This is when data is constantly steaming from primary storage to archive storage. For example, a video surveillance solution may want older video footage to migrate from primary disk to archive tape to free-up primary disk capacity for the latest video footage. For instance, all video content older than 180 days is moved to the tape archive since most analysis of the video is done well within 180 days of an event. This means that every day, a day’s worth of data that is 181 days old is moved to tape, preserving it for later analysis if necessary.

Sizing for your workflow

data archive_ATSB_shutterstock_283470008_03Sept_sized.jpgNow that we have two workflows defined, it’s important to understand the specific user requirements for each when sizing the solution. In a project-based workflow, you need to know the project size and retrieval requirements. You may want to keep the latest project in a “warm” state when you write it to the archive so that a copy is kept on the disk cache for quicker subsequent access. Perhaps you also want enough room to allow projects that are read from tape to stay in the disk cache for quicker access while that project is in demand. Let’s assume your projects are 5TB in size. You want enough room to keep the latest project in a “warm” state and still have enough room to keep two older projects read from tape available for quick access as well. This means you would need 15TB of disk cache. If you treat your archive as more of a “write once, read maybe” then you can get by with a very small cache, which could even be flash. Check out this blog where I talk about how you can use disk and flash storage in your tNAS solution.


In a steady-stream workflow, you most likely only need enough disk to buffer the write operations to tape. This assumes that subsequent reads from the archive would be so that data could be retrieved and written to a separate disk or flash tier for analysis. For example, if someone needed to pull video footage from six months ago, you might use their video surveillance software to identify the specific cameras and times that they want to retrieve. That data would be pulled and staged on some other storage tier for review and analysis as needed. Once again, flash could be used in this situation instead of disk for the tNAS cache.


Sizing your tape library

The simplest way to think about a tape library for archiving is to imagine it as a big bucket with BucketJ.jpgfaucets dumping water into it. The faucets represent the tape drives themselves as they will determine how much data can flow into the bucket at any one time. The size of the bucket represents the media and slots in the library and determines just how much data can be stored. Of course, the kind of drives and media also impacts the amount of data as newer generations of LTO tape can hold more data per cartridge than earlier generations (LTO-6 holds 2.5 TB per tape natively, LTO-5 holds 1.5 TB per tape natively).


In a prroject-based workflow, you may have more time available to move the data so you may not need as many drives or faucets to move your data. A few drives plus a spare drive may be sufficient. In a steady-state workflow, you most likely have a well-defined and possibly shortened timeframe to complete the data transfer. As in our video surveillance example, you only have one day to move all the video feeds from the 181st day before the cycle starts again the very next day. Depending on how much data is created each day, you can determine how many drives are needed to meet the time window. Our testing shows that with LTO-6 drives we can stream up to 120 MB/sec per tape drive in a tNAS solution using RAID 5 disk for cache and 270 MB/sec using flash.  


Sizing the tape library is a pretty straightforward calculation. Like the bucket, you need to know how much data we want to keep in our library for long-term storage. If you want to keep 30 x 5TB projects online in our library, that means we need a library that can hold 150 TB which equates to 60 LTO-6 cartridges (assuming no compression). This easily fits into an HP StoreEver MSL6480 tape library that can be expanded as you add more projects to the archive. Similarly, we can calculate the number of slots needed to store 6 months of surveillance footage by simply determining the data feed rate from each camera and multiplying that by the number of cameras to give us a daily rate of creation. Take that number and multiply it times 180 (6 months) and that’s the total. Keep in mind that you can also create secondary tapes that be exported for off-site storage to protect against disasters such as floods or fires.


When it comes to the right-sized tNAS solution, it’s all about the workflow

Hopefully, you have a better understanding how different workflows impact the sizing of a tNAS solution. It’s not a complicated process but it’s important to understand so your tape archive can provide you the best performance and capacity price point to meet your business needs. Having a strong archive plan can help you better manage your content and maybe even more importantly, your financial budgets.


Learn more

All tNAS white papers

Auto Migration

Big Data


More Around the Storage Block blogs

What is tNAS?

tNAS and Cloud

FLAPE—combining flash and tape

tNAS used at Barrett-Jackson

QStar and HP

0 Kudos
About the Author


Our team of Hewlett Packard Enterprise storage experts helps you to dive deep into relevant infrastructure topics.

Eddie J

"(LTO-6 holds 2.5 GB per tape natively, LTO-5 holds 1.5 GB per tape natively)."


That should be 2.5 TB and 1.5 TB  ;-)


Thanks, Eddie J. Good catch! We made the correction.

-ATSB editor