What implications does this have on widely used security protocols and methodologies?
Tapping and storing terabits of data is effectively trivial, analysis is the hard part. If they hold useful SSL keys (like Facebook, Google, etc) they'll have a much a much easier time figuring out what people are doing.
Assume everything popular is attacked, assume vast libraries of primitive and proprietary cryptanalysis routines.
Of their thousands of hires with security clearance, the cipher newbies are likely given exercises attacking odds-and-ends with massive (old?) Cray orchards.
OTP - is always secure but inadequate for large plaintext. Pad transport is vulnerable.
Naive stream ciphers are seductive but weak.
Brits, MI5(?) had RSA a decade and a half(?) before RSA.
Factorization is RH(`assume Riemann Hypothesis'), perhaps new unpublished vulnerabilities to probable primes?
Better amateur practicing crypto analysts than me can likely offer good points here.
Wikipedia has terrific crypto intro. eg:
http://en.wikipedia.org/wiki/Outline_of_cryptography
and ...