The cloud storage industry has gone through a veritable roller coaster ride during the first few years of its existence. The advantages are clearly there, but a number of factors have caused unease amongst SMB and Enterprise decision makers alike. The cloud security concern has been, and always should be, paramount. Trusting your information to a third party should never be taken lightly, and making a bad decision there can have disastrous effects. The good news is that many of the leaders in the storage space can provide encryption, aggressive SLAs, and physical data center security to make them legitimately trustworthy.
The next concern has always been a lack of standards for the APIs used to access cloud storage. The lack of a standard API means that businesses take a risk by committing to a single storage vendor… their data cannot be transferred to another vendor without great difficulty. This is one form of lock-in, but this problem has also been mitigated with most of the major players moving to a REST-based API. There is however yet another new form of vendor lock-in that businesses contemplating moving to the cloud need to consider…
Latency.
Having your data “in the cloud” can do incredible things for remote worker collaboration, backup, and file sharing, but it is also subject to the major drawback of the Internet… latency. This means that a volume of data that would usually take 30 seconds to upload/download via a SAN may take 10x that, or worse, depending on where the data is stored.
This is not a horrible problem with small transfers, but adds a whole new element of lock-in to an organization that may be trying to store terabytes or petabytes of data in the cloud. This can take days to weeks to transfer via the Internet. Consistent APIs or not, there is a level of commitment that businesses may not have taken into consideration when this is the case! This reality became painfully evident in the shutdown of the EMC Atmos cloud storage service a few months ago. Users were given a short period of time to completely remove all of their data, or it would be deleted.
This is unsettling from several standpoints, but definitely points to the fact that cloud performance itself, and the time that it takes to get data to and from the cloud, is definitely a weakness. Businesses need to undertake the most logical solutions from a cost perspective, a scalability perspective, and a performance/time investment perspective. There are several important things that need to be taken into account with these needs in mind:
1) Where are the storage facilities?
A storage vendor with a globally-distributed data center footprint is always best. The ability to choose where your data is stored should be paramount from a latency, security, and regulatory standpoint.
2) Is it possible to ship a hard drive?
Depending on volume, many vendors are willing to receive a shipped hard drive with the initial bulk of data. This allows the user to have their data uploaded at SAN performance, and be readily accessible immediately.
3) Does the vendor have content acceleration capabilities?
Some storage vendors have content delivery systems in place, and boast advanced solutions for bypassing the public Internet. One great solution is to install an appliance in the user data center to create a high-performance link between the local site and the cloud data center or network. This makes it feel like cloud storage is local.
As always, ask questions, and nail down up-front the prices for these additional services. Latency has been a speed-bump in the development of cloud technologies, but it is by no means an insurmountable obstacle. There is enough industry pressure to drive development of more convenient systems of inter-vendor data transfer.
Jonathan Hoppe
