The untimely demise of Nirvanix has left over 1,000 customers scrambling to migrate data off of the cloud storage service provider and with a short two-week timeframe to save their data. While providers have gone to great lengths to make data import into the cloud easy by eliminating data ingest fees, large data sets in the cloud are difficult to retrieve or migrate to a new target. The recent example with Nirvanix highlights why customers should also consider exit and migration strategies as they formulate their cloud storage deployments.
One of the most significant challenges in cloud storage is related to how difficult it is to move large amounts of data from a cloud. While bandwidth has increased significantly over the years, even over large network links it could take days or even weeks to retrieve terabytes or petabytes of data from a cloud. For example, on a 1 Gbps link, it would take close to 13 days to retrieve 150 TB of data from a cloud storage service over a WAN link.
To minimize risks in cloud storage deployments and facilitate a graceful exit strategy (just in case things go sour), I recommend customers take the following steps:
Leverage Direct Connect. Amazon, Nirvanix, IBM's SoftLayer, and other public cloud storage providers have forged agreements with carrier-neutral data center providers such as CoreSite and Equinix to establish (1 Gbps or 10 Gbps) Ethernet cross connects between public cloud storage and other resources available at the data center site. With direct connect, customers can take advantage of LAN bandwidth within these sites to quickly migrate data to a colocation resource or another cloud, and avoid slow and costly data transfers over WAN links.
Use shuttle services for import/export of large data sets. For payloads that are too large for your WAN link to handle in a timely fashion, some providers have shuttle services, such as Amazon's Import/Export, which allow for speedy data retrieval or import by shipping portable hard drives. While the thought of physically transporting data may seem archaic, a box full of disks shipped overnight will transfer data faster than a WAN link, especially in remote areas where bandwidth is not as easy to attain.
Tap multiple clouds with gateways. Cloud gateways from vendors such as Panzura, Ctera, TwinStrata, and Nasuni have become increasingly popular since they provide an on-premises cache of frequently accessed data to minimize data access requests to a cloud service provider. WAN optimization and data deduplication technologies are also used in these systems to reduce the amount of redundant data that is shipped between a customer and the provider to establish a hybrid cloud environment. In light of the Nirvanix incident, the ability to simultaneously write to multiple cloud storage services is also important since it will allow customers to have copies of strategic content stored in multiple sites in the event of an outage.
A Quick Take report with more in-depth recommendations for cloud storage strategies will be available to Forrester customers tomorrow.