From its beginnings as an alternative to the proprietary DLT format, I must say LTO has been quite a success story. Its history of delivering continuous innovation has been impressive, increasing capacity and performance with every generation, plus adding features such as WORM, encryption and especially LTFS, the Linear Tape File System with LTO-5. Now LTO-6 has expanded the “history buffer” in the compression engine, giving it a 2.5:1 compression ratio and a 6.25 TB per cartridge capacity.

While the roadmap on the LTO Ultrium website has been laid out to Generation 8 with plans for 32 TB of capacity per cartridge, we were told that the next generation beyond that would hold 50 TB. The reason the consortium can keep upping these numbers is that tape as a recording medium isn’t pushing the bit density limits, as disk has been doing for years. Each bit on tape occupies a much larger area than on disk, larger than is required for that bit to be reliably recorded and read back. This means tape can keep packing more bits into the same physical area without killing performance or requiring excessive error correction.

An interesting piece of data we learned during this briefing with the LTO Consortium was that research is showing that while unstructured (file) data is growing much faster than structured (database) data, the majority of that growth from a capacity perspective is in image files, specifically “rich media,” from industries such as media and entertainment, oil and gas, health care, life sciences, as well as general business segments. Pictures, high-resolution graphics and video are pervading the workplace. This means that the types of data that are creating capacity issues for the storage customers you have is also the kind that can’t be readily compressed.

It’s kind of a “perfect storm” scenario for tape. The data sets that are filling up everyone’s disk systems can’t be reduced with the deduplication and compression technologies that disk has been relying on for years to maintain its economics. As a VAR/MSP, this means that your calling base may finally be open to a meeting to learn about active archives (and active archiving software) or simply using LTFS to create a 6 TB portable storage container.

Another area of discussion was around backups, data corruption and the fact that a lot of newer IT folks have never actually used tape. I’ve been noticing an increase lately in the number of phishing emails I’m getting using names like Amex, ACH, the IRS and others. All it takes is for one person in an organization to click on one of those attachments and all kinds of problems can occur. With disk-based backups, data corruption can propagate throughout an environment before anyone realizes they have a problem.

Tape provides a physical copy of the company’s data set that won’t be overwritten with a corrupted replica, as can too easily happen with disk. This “backstop” data protection is very inexpensive, especially with the capacities of LTO-6. For the IT admins who’ve not been introduced to tape, this may be a real opportunity — cheap insurance, to say the least.

Follow me on Twitter: EricSSwiss

Successful vendors know this and take pains to develop these tools. The reference data and archive space is a great example. When an organization is unhappy with its existing archive solution but is faced with how to move hundreds of terabytes or petabytes of files from one platform to another, they may just grin and bear it. They’re less likely to trust a VAR who comes in the door selling another solution if it doesn’t have a reasonable data migration strategy. Quantum seems to have been listening to its VARs on this one.

The StorNext File System (SNFS) and Archive Manager solution now has a module called the Archive Conversion Utility (ACU). It can provide a sane migration path for an organization ready to leave its Sun/Oracle SAM-FS or QFS system, one that won’t end up requiring a month of PS time, six months of nights and weekends by the IT team or extended periods of degraded performance for the users. By importing the legacy file system metadata into the SNFS namespace, all legacy files can be accessed from within the StorNext file system. Then, when files are saved, they go directly into the SNFS namespace. Later, administrators can schedule the conversion of archived files that aren’t being accessed over to the SNFS, at convenient and nondisruptive times. In this way, the transition can take as long as necessary without interrupting data access for users or creating an after-hours hardship for IT administrators.

While Quantum certainly markets its own libraries for use with StorNext, the SNFS is OS-agnostic and supports most major tape libraries. This flexibility, along with the ACU module, makes StorNext an attractive solution for VARs.

Follow me on Twitter: EricSSwiss

To be fair, backup has “evolved” into an application that moves an entire organization’s information assets from a large number of random access devices to a smaller number of serial devices (in the case of tape), over a network in real time. A big job indeed, which is the point.

As a wise old storage engineer used to tell me, “The only way to improve backup is to do less of it.” So let’s improve your customers’ backups. The concept of archiving certainly isn’t new (see HSM, ILM, DLM, etc.), but it’s recently become a viable alternate landing place for primary data, and it’s actually a better one for long-term retention.

Enterprise archive systems from companies like EMC, Nexsan and Permabit, for example, provide a Tier 1-class destination for data that’s not changing or is not accessed often. They’re built with a high-availability and redundancy architecture that users now expect, but also add data reliability and continuous verification features needed for long-term viability. There’s also WORM and encryption available to meet security requirements.

As a VAR, your job is to solve problems for your customers. True solutions to these problems are seldom “point solutions.” Let’s face it, if this stuff was easy, your customers wouldn’t need you. Implementing an enterprise archive will provide real relief for your customers’ backup issues, relief they can’t get with deduplication. And, it will set them up to enjoy other real benefits (through their VAR) as they get more of their enterprise’s data under control for the long term.

After all, as a VAR, solving your customers’ problems is the way to the promised land — becoming their “trusted storage advisor.” Attacking your customers’ issues leads to that messy process of problem definition, proof of concept, system design, product evaluation, implementation, testing — you know, integration and professional services engagements. It’s the “value” in “value-added reseller.” 

Back to dedupe …

So, dedupe does generate POs, but does it really solve any storage problems for the customer? Well, it doesn’t reduce primary storage requirements. And, as mentioned in recent articles on Storage Switzerland, it doesn’t reduce network requirements or administration requirements. It also doesn’t usually reduce backup windows (for target-side dedupe, anyway).

What dedupe provided when it first appeared several years ago was a much simpler solution for disk backup than virtual tape libraries (VTLs). It didn’t really solve the fundamental backup problem then, and it still doesn’t now. 

Dedupe does make backup more efficient by reducing the amount of redundant data that has to be physically written in a backup. This enables you to postpone the purchase of additional backup-to-disk hardware. (I know dedupe appliances can also provide a relatively simple remote DR solution, but let’s pretend we’re only interested in local backup.) But it doesn’t really reduce the number of files in primary storage that need to be handled by the backup system, by its associated networks and servers or by the backup administrator.

For most organizations, the way to do that is with an archive. When files are moved off to an archive, they’re out of the backup stream altogether and the amount of data handled, processed and backed up is reduced. This also reduces per-terabyte storage acquisition costs and all the other costs associated with a larger primary data set. This is the kind of real problem solving that will keep your customers for the long term — and generates orders along the way.

Now, we’re not talking about your grandfather’s HSM here, but enterprise archive. More on this next time.

Eric Slack is a senior analyst for Storage Switzerland. He can be reached at eslack@storage-switzerland.com.