Posted on March 24th, 2009 4 comments
Cloud storage is getting a lot of attention these days, and not all of it is good. Over the years, lots of money has been plowed into building a business model in subscription-based on-line storage. Many of these companies have already gone bust or have been acquired. Just recently there was an article that Carbonite, a leader in online storage, is suing a hardware provider for losing customer data.
A good backup strategy requires a multi-pronged approach, not a single “silver bullet” solution as cloud storage is sometimes thought to be. People who do backup for a living understand this, yet many consumers are just beginning to become aware of a need to keep irreplaceable data like digital photos safe from computer viruses, hardware failures, fire, theft, and natural disasters. A major attraction of cloud storage is that presumably the data is stored on very reliable hardware and looked after by IT professionals and copied to redundant, geographically distributed data centers. But if you’re happily storing your data in some amorphous data cloud, if any data goes missing, you might be the only one likely to notice. That may happen quickly if it is data that is accessed frequently, such as data on an active website, but in the case of an archive of digital photos, if you don’t bother to check them periodically it could be a few years before you might notice some or all of the photos are gone (most likely while you’re attempting to do a restore). If that happens, what are the odds of getting your data back? I’d say not very good, since backup media like tape in data centers is frequently re-cycled and overwritten. Sure, there may be an archive copy off-site in a box of tapes, but I’ve been down that road and I know where it leads, and it’s usually not to your lost data.
What happens if the cloud storage company isn’t able to cover its expenses and turns into a money loser for its investors, as has happened several times already? Odds are, they will cut and run, telling you in a hasty email that your data is there for the downloading. You have a month to get it, otherwise it’s gone forever. That’s what HP’s Upline service told its customers recently.
I don’t think that cloud storage is necessarily a bad idea, it’s just that it’s hard to make money with it because people don’t like paying perpetual monthly fees for a benefit that is largely invisible to them. Companies offering a service love recurring revenue models, and will go to extreme measures to create one, not unlike a crack dealer giving away the first hit for free. People do crazy things for crack, and companies do crazy things when they hear those magic words as if sung by angels, ‘recurring revenue stream’.
I was comparing the price of Amazon’s S3 service to using a web hosting account like GoDaddy as one’s personal cloud storage. For 150GB of storage and 1.5 TB of monthly throughput, GoDaddy charges $7/month. There are many other providers with similar hosting plans. For Amazon’s S3, it costs $.15/GB per month of storage + .10/GB bandwidth charge for uploading and $.17/GB for bandwidth charge for downloading. It sounds cheap, but if you already have a hosted website you may have about 145 GB of storage left over, and thus the web host solution is free whereas the Amazon S3 solution would set you back another $22/month for 145 GB of storage. And you would need to spend about $40 extra for the bandwidth to upload and download the data just once.
Moving 145 GB of data on a typical broadband connection with a capped upload speed of 1 Mbps would take about 300 hours (~13 days) of continuous uploading. That’s another downside to cloud computing. The data transfer times are usually deal breakers, if one bothers to do the math. Would you wait 2 weeks for a backup to complete? I sure wouldn’t. Even to restore that amount of data on a respectable 3 Mbps download connection would take more than 4 days. Again, not a very compelling solution.
There is a place for cloud storage, but it’s probably best done in small amounts, perhaps a few GB of really important digital photos, and in those cases, a simple regularly-scheduled upload to a web host would be the most economical way to go.