Archiving: Tape rises from the dead - Part 2

In the previous column, key issues regarding backup and archive solutions were reviewed. Each approach represents quite different solutions to non-similar tasks. A backup solution represents a business insurance policy. It helps the company continue operation over days, weeks, perhaps months after the occurrence of some catastrophic event. The data for a backup solution should be stored off-site, perhaps hundreds of miles distant.

An archive solution is targeted at retaining an ever-growing amount of content and providing effective management over that content. An archive must manage live (and perhaps dead) and ever-changing content. It must delete content when rights expire. It should seamlessly move content between storage technologies as users require access. And, an archive solution needs to be an evolving tool, one that can adapt as formats and equipment improve.

Mentioned earlier, a backup solution is an insurance policy — a tool that enables your business (a broadcaster or production company) to continue operation should a physical disaster hit your location. Typically, the solution moves the content you need to operate for a brief period of time to some remote location. You then either operate from that remote location or stream back that data to stay on the air — or in the production business.

Table 1 (courtesy SGI) illustrates some key differences in backup and archive systems. (Click on table to enlarge.)

With this background, let’s look closely at the issue of spinning disks versus tape. Which is better? That is the wrong question. The better question may be: Which technology is better for what application?

I’ve heard engineers claim that disk is preferable over tape. These professionals then offer a host of reasons why disk is better than tape: faster, easier to manage, less expensive, ever-increasing storage density, etc. Besides, tape is dead, right?

Maybe not.

To begin comparing disk and tape solutions, let’s start by looking closer at the cost of each storage platform.

Cost of storage

A research report from the University of California, Santa Cruz, summarized business and media storage costs. The study said that 90 percent of data stored was never again accessed. Less than one percent was accessed more than five times. See Figure 1. (Click on figure to enlarge.)

What is it costing you to store 90 percent of your media that may never again be used? The only acceptable answer is free. If it costs anything to store that 90 percent of your data, the costs may be too high.

In a report from the Clipper Group, the writers said, “Storing infrequently-accessed data on disk is equivalent to keeping your car running in the driveway – it wastes energy and it costs money.” Power consumption should be a part of TCO storage calculations.

Echoing earlier studies, analysts David Reine and Mike Kahn of The Clipper Group published a 12-year TCO archive study comparing a SATA disk system with an LTO-5 tape library system. The paper said that the average spinning disk system TCO was 15 times more than a comparable tape system. In addition, disk solutions used 238 times more energy than did tape solutions. In this study, costs to just power the spinning disks over the 12-year period was more than the total tape system purchase cost. See Figure 2. (Click on figure to enlarge.)

There are other kinds of storage costs: equipment acquisition, maintenance, replacement and power costs. Acquisition costs are determined when you first buy the gear. Ongoing costs are less controllable and harder to identify. Because government will increasingly demand cleaner (read that as more expensive) power generation, electrical costs will increase. Bottom line: Going forward, the costs of “powering” storage will become an ever larger component of storage costs. Therefore, when making storage decisions, also consider power costs.

Not yet convinced?

Table 2 compares the cost to store 1PB of data on both disk and tape platforms over a five-year period. (Click on table to enlarge.) The estimates include both CapEX and OpEX for that period.

The tape library had 24 LTO-4 tape drives and 800 cartridges. The spinning disk solution consisted of 1TB SATA drives configured as RAID 6. Measured over five years, spinning disks cost 3.6 times more than a tape-based solution.

According to the authors, typical storage CapEX runs between 20 percent and 40 percent of total TCO. OpEX expenses are more difficult to calculate, but your financial people can help. OpEX charges may include the cost of people, power, cooling and warranties.

The next column will examine a newer alternative to strictly tape or disk storage. Several vendors are offering what they call Active Archive. This blend of technology hopes to offer the best of both worlds. We’ll look at specific solutions, performance and standards.

CATEGORIES