Deciphering the secrets of network encryption
Focus on all-round security
Modern data centres are pools of shared resources and segmenting the data in them has become a requirement as much for regulatory as for security reasons. It is no longer enough simply to provide access and expect individual users to take care of security themselves.
Relying on file system permissions or even virtual provisioning is inadequate. The larger the datacentre, the more critical that end-to-end security, with a focus on encryption, becomes.
Encrypting websites is now mainstream, though the system requires major revision. Less popular – but just as important in shared environments – are the back-end encryption systems available to large data centres.
Tower of Babel
Modern filers offer encryption. Network switches offer encryption. But making sure that all of the various widgets can talk to each other is where everything becomes a little sticky: past a certain point, manually managing encryption keys is simply not feasible.
Keeping track of your encryption keys is the job of an enterprise key management system (EKMS).
Enterprise players in many markets each have their own solution to this problem. The two most popular EKMS offerings available are from RSA and NetApp. To avoid key management sprawl, other encrypted devices must be capable of working with these systems.
Key management interoperability is vital and standards are important for any large-scale encryption project to succeed. The Oasis Key Management Interoperability Protocol (KMIP) is one worthy initiative in this space, allowing servers, storage, switches and other devices created by different manufacturers to operate effectively in a unified environment.
NetApp is another involved vendor, offering storage filers as well as key management software. Assemble the various pieces, and you can achieve a robust encryption environment.
Data on the filer is encrypted. Even if somebody walks away with the disks, they would have little chance of interpreting anything intelligible without access to the key management system and the appropriate permissions.
Data being written to tape can be encrypted. This is important as backup tapes are often overlooked as a potential security breach.
The transmission of data between the filer, the switch and the end HBA can all be encrypted. It is not inconceivable that attempts will be made to sniff information on the wire and the right algorithms can make decrypting this information without detection nearly impossible.
You would be unlikely to sneak that much gear into somebody’s data centre without being noticed
Today, AES-256 should be considered the minimum security standard employed on any network. Though theoretically it is a vulnerable algorithm, the equipment required to attack AES-256 is still measured in acres.
You would be unlikely to sneak that much gear into somebody’s data centre, or park it outside, without being noticed. Nor is there any chance of being able to attack AES-256 remotely: the bandwidth requirements would be extraordinary.
Encryption and segmentation of data at the network level is an expected security element of modern large-scale networks. On its own, the security it provides is not enough. Yet neither do any other standard security precautions provide a complete solution.
Network encryption is one piece of the puzzle whose various pieces have to be tracked and managed throughout their lifetime. ®
Erm, you mean customer lock-in ?
Sorry for my verbosity.
Technically good but from a cost standpoint another matter
I thought this was a good technical marketing speak, that is to state the obvious technical merrits but lacks a resolution to the problems that Data Centers have to deal with such as cost. Data Centers are coin operated and current solution offerings are way to expensive to implement into a Data Center or the Cloud as a one to one model, it has to scale to a one to many offering. This is and will continue to be a big issue for Data Centers until security vendors change to a licensing model that allows for a data center to 1) be able to resell the services to it's customers at a much lower cost average. Current per seat cost is about $60.00 for encryption. 2) Change to a Data Center Specific type of a license to be able to sell this to a Data Center under a subscription service offering. These two simple steps creates a long term committment for both the customer and vendor, it also creates a long term re-occuring revenue stream and stream lines sales to leverage the Data Center sales team who already owns the relationship and as a Subscription license agreement it creates a capital expense for the data center ie they get to right the entire thing off at the end of the year. Security for Data Centers is not the problem it's the cost vendors are wanting to charge them under an old way of doing things when Cloud computing was not available.