To deal with the challenges of both computation-complexity and algorithm-scalability posed to the design of an IPSec engine, we develop PAC (parallel algorithm core), called PAC, employed in an IPSec engine, which can meet requirements of both exploiting parallelism existing in IPSec packets and offering scalability in both the scales and types of cryptographic algorithms. With three kinds of parallelism and two kinds of transparency defined, a novel hierarchy of the specifically-designed parallel structure for PAC is presented, followed by corresponding mechanisms. With a simulation, the scalability of PAC is examined. For the purpose of performance evaluation, a Quasi Birth-and-Death (QBD) process is then established to model a simplified version of the proposed PAC. Performance evaluation of PAC in terms of two representative measures, throughput and mean packet waiting time, is numerically investigated. A comparison study is done on a simulation basis. Conclusions are finally drawn for providing a helpful guideline for both the design and implementation of our proposal. To deal with the challenges of both computation-complexity and algorithm-scalability posed to the design of an IPSec engine, we develop PAC (parallel algorithm core), called PAC, employed in an IPSec engine, which can meet requirements of both exploiting parallelism existing in IPSec packets and offering scalability in both the scales and types of cryptographic algorithms. With three kinds of parallelism and two kinds of transparency defined, a novel hierarchy of the specifically-designed parallel structure for PAC is presented, followed by corresponding mechanisms. With a simulation, the scalability of PAC is examined. For the purpose of performance evaluation, a Quasi Birth-and-Death (QBD) process is then established to model a simplified version of the proposed PAC. Performance evaluation of PAC in terms of two representative measures , throughput and mean packet waiting time, is numerically investigated. A comparison study is done on a simulation basis. Conclusions are finally drawn for providing a helpful guideline for both the design and implementation of our proposal.