1. strong fundamentals of L2/L3 networking
2. software defined networking state of the art
3. linux and windows virtualization state of the art
4. distributed storage state of the art
5. familiarity with commodity 64 bit hardware of all varieties (supermicro, dell, hp, etc.)
6. kernel debugging skills (maybe not development)
7. networking debugging skills @ the packet level
8. strong understanding of http and proxying
9. strong understanding of load balancing
10. strong understanding of wide area networks
11. physical datacenter knowledge i.e. how to actually install shit and get it to work together.
12. power fundamentals
for example, solve this problem:
we have a datacenter in oregon and a datacenter in virginia. connect them physically and logically. know what the goals are, know who to call, what to ask for/buy, and how to debug it.
1. Formal specifications. Complex distributed algorithms are too difficult to understand without exhaustively checking a design for all possible executions that end in dead-locks or violation of some invariant. Clouds require sometimes complex orchestration between various services. Unit tests aren't going to cover you if your design isn't correct (they'll ensure that some small portion of your system is correct and scale has a funny way of revealing everything you didn't test for).
it was implied in point 3, but yes, and (implied in point 5)... know how to do this for physical servers too. i.e. the machines that are actually doing the computing that everyone thinks is happening 'in the cloud'.
i.e., how do you assembly-line build an entire rack or cage without manually installing any software? it's not easy. in fact it's very hard, probably even the most advanced groups require some manual intervention or at least some manual troubleshooting every once in a while.
1. strong fundamentals of L2/L3 networking 2. software defined networking state of the art 3. linux and windows virtualization state of the art 4. distributed storage state of the art 5. familiarity with commodity 64 bit hardware of all varieties (supermicro, dell, hp, etc.) 6. kernel debugging skills (maybe not development) 7. networking debugging skills @ the packet level 8. strong understanding of http and proxying 9. strong understanding of load balancing 10. strong understanding of wide area networks 11. physical datacenter knowledge i.e. how to actually install shit and get it to work together. 12. power fundamentals
for example, solve this problem:
we have a datacenter in oregon and a datacenter in virginia. connect them physically and logically. know what the goals are, know who to call, what to ask for/buy, and how to debug it.