What’s Cheaper Candles or Light Bulbs?
By David A. Chapa, Chief Technology Evangelist, office of the CTO, EVault
I recently read a commentary published on the Computerworld website entitled “Tape versus disk: The backup war exposed” (Apr 29, 2013), which caused me to shake my head and mumble to myself, “here we go again”.
You see, at the beginning of this piece, the author makes a point that most CFOs view backup and disaster recovery programs as just insurance policies, so the least expensive one is what they select. For this reason I believe the writer wrote from the perspective of cost/performance comparison of disk and tape.
I think his supposition regarding CFOs may have been true in the 1980s but not now. Today’s CFOs are more technologically savvy and increasingly engaged with their CIO colleagues. CFOs, good ones, know the difference between price and value, price is what you pay whereas value is what you get.
For this reason and others I stopped mumbling to myself and put pen to paper, which led to this response because I’m as tired as the next person at hearing, reading or listening to blogs, articles and lectures that compare tape to disk. So to begin the last narrative you’ll ever need about this subject I’ve collated my nearly 30 years of industry experience and know-how to offer the reader a perspective that I’m confident you’ve not heard before – a comparison of tape and disk from the perspective of candles and light bulbs.
Is tape cheaper? Probably. Are candles cheaper than light bulbs and electricity? Again from an acquisition perspective, probably but I don’t see a mad rush of folks running out to stockpile big candles to light their homes, offices and restaurants.
Have candles died or lost their usefulness? Absolutely NOT! Where would birthdays or romantic dinners be without them? And what about creating that special ambiance when the power goes out? My point, here, is that comparing these two mediums, purely on price alone makes for a ridiculous argument.
It is not just about cost, it’s also about application and what makes the most sense from an efficiency perspective. Before going further let me provide a little background, since I did spend a good part of my career in IT as a backup admin.
In my experience, disk has for a long, long time been used as a backup medium – maybe as long as tape itself.
- In a role I had back in the late-80s I managed backup operations writing to 288MB disk packs and a series of 9-track tapes on a reel to reel drive, both disk and tape.
- In the mid-90s I worked for a market leader in backup and recovery software targeting the UNIX market. We sold two flavors of the solution, a basic config that backed up to disk or an advanced configuration that supported both disk and tape.
- Enter Y2K and the need for speed. Tape drives could not keep up and backup windows were being missed, so we multiplexed or interleaved data onto tape from multiple sources to take advantage of streaming. Where we failed was on the recovery side. It took 2-5x as long to recover a system when its backup was intermixed with other client data.
It was about this time that I started to see a shift happen in IT environments. Tape was being moved further to the right of the technology timeline, making room for disk.
Disk, being a random access medium was quite versatile for the backup admin. She could backup the required data to disk then begin making copies of that data to tape for offsite storage. If during this copy process a restore request came in for data within the same set being copied to tape it was no problem. Again a random access medium is just that – random. She could satisfy both requests simultaneously. You cannot do that with tape – a sequential medium.
The problem with using disk for 100% of your backup is the need to have enough storage to protect your active data. This presented a big budgetary problem for IT until deduplication entered the fray and offered IT an opportunity to backup to disk without needing to purchase 1:1 storage. Deduplication would advertise 10:1, 15:1 or 20:1 in capacity optimized storage.
Between 1997 and 2003 I worked as a data protection consultant working with end users, solution providers, and vendors, to help identify where the market was heading and what features were needed to stay ahead of it. I always stressed the importance of using the right technology for the right purpose; hence my aforementioned candles v. light bulb and defense of today’s CFOs references.
According to Jason Buffington, a Senior Analyst at Enterprise Strategy Group, (ESG) and author of “Data Protection for Virtual Data Centers”, the best piece of advice he received around this topic was at a Gartner CFO conference from someone who had attended the ROI breakout. “…if a significant proposal was submitted for review and it had a TCO projection and ROI analysis on its first submission, it would be approved over 40% more often than those that did not have those calculations.”
He makes very good points in his book around TCO and ROI. In particular is his methodology for calculating ROI, which I have used, myself, for many years. Very simply, as Buffington states, ROI is a comparison of one’s BIA (business impact analysis – what does it COST to do nothing to problem X) to TCO (what it costs to solve problem X).
According to a 2013 CW Forecast Survey, 64% of the respondents expect to make a major IT investment in the next 12 months, so I would suppose these IT executives won’t be looking for the most inexpensive solution but rather one with the greatest value to the organization.
Through its ‘Research Report: 2013 IT Spending Intentions Survey’, ESG also reveals the top three business initiatives to impact IT spend in the next 12 months:
- Cost Reduction Initiatives: 43%
- Business Process Improvement: 31%
- Security/Risk Management Initiatives: 31%
Admittedly, this is the long-way of saying that I disagree with the (Computerworld) author’s depiction of CFOs today. This “backup war” that he refers to is much greater than the most inexpensive solution; it is about real solution value.
Question: So, what is the value you need/expect from your solution?
Answer this and I believe you will be well on your way to identifying the best fit for your organization.
I could spend a lot time addressing this issue with charts and graphs illustrating how disk compares to tape but to accelerate the process one needs only to ask themselves the “price v. value” question, which is the very same question CFOs will ask when a proposal, to upgrade the data protection solution, lands on her desk.
Most customers I talk to, today, express the need for fast recovery in the event a file is deleted or corrupted. The other expressed need is for a Disaster Recovery (DR) copy to be sent offsite as soon as possible. I suggest you follow up, as I always do, by asking the customer to define what ‘fast’ and ‘as soon as possible’ mean. Having this additional information will allow you to better understand the value expectations.
Back in the late 80s and early 90s, ‘as soon as possible’ meant the next day, morning preferably, whereas ‘fast’ meant within an hour or two. At that time, many ROI decisions were based on the methodology, referenced earlier in this piece. I suggest you follow up, as I always do, by asking the customer to define what ‘fast’ and ‘as soon as possible’ mean. Having this additional information will allow you to better understand the value expectations. In some cases, based on the ROI, some customers resolved to a 24 hour offsite window for the DR copy and a one or two hour recovery time objective with a 24 hour recovery point objective.
Today, however, it makes perfectly good financial sense to look at capacity optimized disk based backup solutions with replication to solve problems that most customers are facing with fast recovery and offsite copy as soon as possible. In fact, many customers are now looking at cloud services to help offset costs and handling for its data protection strategy and solution. ESG reports 71% of customers in a recent survey indicated an increase in budget spend on cloud services for 2013.
What I don’t understand in this Computerworld commentary is why the author’s comparison of Tape to Virtual Tape Libraries (VTL – disk presented as tape drives, cartridges and tape library automation). This is not a good comparison if you take the title of the piece seriously.
When you choose backup to disk you eliminate many of the additional costs associated with tape, (virtual or physical), such as licenses needed to pay to your favorite backup vendor. When you choose a capacity optimized disk solution that is also Cloud Connected you eliminate the additional power and cooling required onsite (smaller onsite disk repository for the most recent data required for recovery). You further reduce the management overhead and increase overall operational efficiency. For growing businesses, a software as a service, (SaaS) model for data protection may be a better overall option.
In his closing remarks the author of the Computerworld piece makes another point regarding standalone drives. He’s “sure the CFO will not care that someone needs to be onsite all weekend managing tapes”, to which I reply bullocks! Based on my first hand experience with C-level executives, the research from ESG and the survey cited from Computerworld, I conclude that CFOs do care. Having someone spend a weekend flipping tapes does not make for an improved business process for IT.
In the end don’t take my word for it – take a serious look at the price v. value of disk and tape to see if the solution will meet your expectations, because ultimately it is about what is best for you, your organization and your requirements.
Now, if you’ll excuse me, there’s a storm heading my way so I need to flip on the light switch and find those candles I put away in case we lose power.
