1. robustness | minor breakage is hidden and recovered internally |
2. redundancy | immediate failover to available spare connections |
3. reconnection | persistent attempts to resume a broken connection |
4. bandwidth | multiple simultaneous routing across different NICs |
5. security | openSSL-based encryption of the connections |
6. authentication | openSSL third-party certification |
7. compression | not implemented yet |
8. size limits | 32bit kernel and/or architecture-imposed limits circumscribed |
9. reporting | /proc-based interface |
10. tuning | some support via client signals and module parameters |
11. internals | kernel requests are aggregated, as in IDE and floppy devices |