Multi-Pathed TCP/IP (MPTCP) and I/O

By Henry Newman

In light of the recent Black Hat conference, I thought it might be interesting to read up on how storage might be impacted by network communications.  There is an interesting talk is planned at Black Hat on MPTCP. You might ask: does this matter for storage? I think that the answer is yes. And the reason is that as we move forward with storage systems and data access, more of it will be done via direct access over TCP/IP and various IP based protocols.  This means that the security issues mentioned in the article are going to be a problem all the way from your cell phone to the IP-enabled disk drives.  I think some of the problems could be solved, but they require some big changes.  Some fixes:

1. We all need to move away from IPv4 to IPv6. Of course this in itself is not going to solve the problem but with IPv6 comes IPsec and we might as well make it far more difficult to sniff open text packages. This is not to say with enough computer power the packets could not be cracked, but why make it easy?

2. Therefore, of course, we need IPv6 chipsets with IPsec offload on our cell phones. This is not trivial nor will it come without a cost.

IPv6 is ready to go except that the infrastructure is not really here.  We keep hearing over and over again about running out of IPv4 addresses, but without a commitment from the major internet backbone players we have a chicken and egg scenario.  We all need far faster data access, and MPTCP can bring us that. But we all surely need greater security and it is clear that we need to have IPv6 and IPsec as part of our digital lives.

Photo courtesy of Shutterstock.

This article was originally published on August 27, 2014