What happens to your data when a cloud vendor goes out of business?
That is likely the question being asked by customers now that Nirvanix is leaving the market. It has been reported that the UK vendor and partner Aorta Cloud, a British company, is working to take over Nirvanix.
The question I believe worth asking is what happens to your data if you are a customer, and this is just one of many that should be asked, given this might be one of the first of the bigger vendors to leave the market, but surely will not be the last.
Nirvanix gave their customers till the end of September to get their data out. So this situation begs the question, if a vendor goes belly up, what happens to your data and how do you get your data that might be business critical? Other smaller cloud vendors might be at risk, leaving us with only with the big three: Amazon, Google and Microsoft.
But back to what happens with your data. If the lights are turned off, do have to become a creditor in bankruptcy court? Do you have to hire an attorney so other creditors cannot get your data as part of their settlement? I am not an attorney so I do not know, but these are the types of random things that I think about.
Clearly, I would want the data encrypted, and I would maintain my own encryption keys because that way even if someone got the data it would not do them any good. What legal precedents exist in this area for judges to draw upon? I did some searching and did not find any, but these are the types of questions that need to be asked so people will know the outcome in advance.
This unfortunate incident clearly points to why I never recommend putting business critical data into the cloud. I recommend managing your own keys, until the legal issues are settled.
Labels: data storage, cloud computing, Cloud Storage, encryption
posted by: Henry Newman
Maybe TDK knew what was coming when they got out of the LTO (Linear Tape Open) media market. Oracle just recently announced the T10000D tape drive with 8.5 TB of capacity (note that is 3.4x greater than LTO-6). For, say, a 10 PB archive with LTO-6 that is 4000 tapes for LTO-6 and 1177 tapes for T10000D.
We can also assume based on history that IBM will be having an announcement in the next few months on their new enterprise drive. Based on industry speculation, both IBM and Oracle are expected to have similar capacities, so LTO-6 will be left in the dust in terms of density and performance. The increased density also translates into reduced floor space. For most companies this has a significant cost in the computer room and there is of course the cost of each slot in the library and potentially software licenses.
The bottom line, as I have been saying, is that that the cost difference between enterprise tape and LTO is not as much as people think. And with enterprise tape there are many advantages in terms of drive design and cartridge construction, etc.
So what I thinks this all means is that LTO technology is going to have significant pressure placed upon it to grow quickly. The problem is that one company is not responsible. But a group of companies – both tape drive vendors and media providers – have to work in lockstep to come up with an agreed upon direction and step of products that interoperate.
While this might have been a good idea 10 years ago, I think there are more technical challenges and material science issues to cover than there has been in the past. Both IBM and Oracle are using barium ferrite tape (BaFe) made by FujiFilm for their enterprise, while all media vendors but FujiFilm use older metal particle (MP) for LTO-6.
Labels: data storage, LTO
posted by: Henry Newman
Before I start I want to clearly state that I think that tape is a required data storage technology today. And although LTO (Linear Tape Open) is not as reliable, as dense or as fast as enterprise tape, it would be not be a good thing to lose the technology.
Now that TDK has announced they are getting out of the LTO market, leaving three manufacturers of tape and one distributor left, it begs the question: will the LTO consortium survive?
Of course we have three manufacturers of tape drives, but is that not a bit iffy also? Quantum has well-known financial issues and recently outsourced the library development. HP storage revenue has dramatically dropped. Who knows what IBM really wants, as they make both LTO and enterprise tape.
The last generation of LTO had a less than expected density improvement and the performance increases have not been very noteworthy over the last few generations. While the cost of LTO tape is less than enterprise tape, LTO tape is not as reliable from the hard error perspective and cartridge design. LTO is not a dense, so it requires more library slots for the same amount of storage. LTO is not as fast, so it requires more tape drives, which takes library slots.
Some of the recent studies I have done for customers suggest that the cost of LTO and the cost of enterprise tape for 5 PB and greater configurations are not that much different. Which leads to the question today: If LTO were to go away, would it make much difference to the tape community other than the initial problems that would require people to migrate to new technologies?
This of course would cause significant harm to smaller tape library companies, which have no relationships with enterprise tape vendors. Nothing is going to happen immediately, but it is something that should be followed closely if you are invested in LTO.
Labels: data storage, tape storage, LTO
posted by: Henry Newman