Open Forum 2015 11th November 2015
SIP Overload Control (SOC) Task Group Philip Williams – BT Chair of Group
Drivers SIP increasing Wider application Protocol efficiency connectivity applications traffic Wider application not just session control Protocol efficiency verbose? Generic overload control principles still apply
Overload control imperative Excessive nodal load high response times timeouts + caller abandons + retries ineffective workload ‘goodput’ crash Nodal Excessive destination focused load high busy rate ineffective workload ‘goodput’ reduction Destination
SOC TG Strategy Basis – 3 IETF SOC RFCs Requirements Options Guidance need for external restriction of load demonstrated prior experience Requirements clarify: purpose, scope of application integration with other functions / signalling protocols overcome any RFC design issues or gaps traffic priorities Options identify what should be optional and what shouldn’t Guidance implementation - fill in the gaps highlight re-use opportunities Recommendations the detail
Progress – RFCs Well understood choices highlighted some issues identified Title RFC Purpose Method SIP Overload Control 7339 Nodal overload control Hop-by-hop (Via Header) SIP Rate Control 7415 " SIP Load Control Event Package 7200 Destination overload control Subscribe-Notify
Progress - Requirements Source (Client) SIP message restriction algorithms Max Rate – ‘rate-based’ or Proportional rejection – ‘loss-based’ Interaction with load-balancing & failover Load-share → Restrict (SOC) Restrict (SOC) → Load-share ? Differentiated workload - message/session priorities Target Server, Source Client: Which messages should they count? Mixed protocol architecture, e.g. ISUP, BICC ISUP AACC SOC hop-by-hop Non-conforming traffic sources Can they be detected? What action should be taken?
To be continued … Would you like to be involved?