Method, apparatus and system for handling node failure

A method for handling node failure, comprising:

• receiving, by a mobility management entity (MME), a context processing request message carrying a packet data network connection set identifier CSID mask of a packet data gateway (PGW), a first match word, an internet protocol (IP) address of the PGW and an IP address of a serving gateway (SGW) from the SGW when the SGW detects that the PGW has restarted;
• obtaining, by the MME, a second match word, according to the CSID mask of the PGW and a CSID of the PGW in context information corresponding to the IP address of the PGW and the IP address of the SGW stored in the MME;
• and
• deleting, by the MME, context information when the first match word matches the second match word in the MME.

A serving gateway (SGW), comprising:

• a processor, configured to detect that a packet data gateway (PGW) has restarted and obtain a first match word according to a packet data network connection set identifier (CSID) mask of the PGW and a CSID of the PGW in context information associated with the PGW stored in the SGW;
• and
• a transmitter, configured to send a context processing request message carrying the CSID mask of the PGW, the first match word, an internet protocol (IP) address of the PGW and an IP address of the SGW to a mobility management entity (MME);
• wherein the MME obtains a second match word, according to the CSID mask of the PGW and a CSID of the PGW in context information corresponding to the IP address of the PGW and the IP address of the SGW stored in the MME, and the MME deletes context information when the first match word matches the second match word in the MME.