The paper introduces an Information Dispersal Algorithm (IDA) that breaks a file into \( n \) pieces, each of length \( L/m \), such that any \( m \) pieces are sufficient to reconstruct the original file. The IDA is designed to be computationally and space-efficient, with the total number of pieces being \( (n/m) \cdot L \). The algorithm is applicable to secure and reliable storage, fault-tolerant transmission, and communication in parallel computers. The paper discusses the security and reliability of the IDA, including its use in storing and transmitting files in distributed systems, and its application to routing data in parallel computers. The routing algorithm is shown to achieve fast transmission, small buffer sizes, and high fault tolerance, with the probability of all packets arriving at their destinations being at least \( 1 - N^{-4} \) and the probability of packet loss due to link failures being at least \( 1 - N^{-\alpha} \), where \( \alpha \sim 0.25 \cdot \log_2 n \). The paper also introduces a simplified routing method that avoids costly route computations and achieves similar performance.The paper introduces an Information Dispersal Algorithm (IDA) that breaks a file into \( n \) pieces, each of length \( L/m \), such that any \( m \) pieces are sufficient to reconstruct the original file. The IDA is designed to be computationally and space-efficient, with the total number of pieces being \( (n/m) \cdot L \). The algorithm is applicable to secure and reliable storage, fault-tolerant transmission, and communication in parallel computers. The paper discusses the security and reliability of the IDA, including its use in storing and transmitting files in distributed systems, and its application to routing data in parallel computers. The routing algorithm is shown to achieve fast transmission, small buffer sizes, and high fault tolerance, with the probability of all packets arriving at their destinations being at least \( 1 - N^{-4} \) and the probability of packet loss due to link failures being at least \( 1 - N^{-\alpha} \), where \( \alpha \sim 0.25 \cdot \log_2 n \). The paper also introduces a simplified routing method that avoids costly route computations and achieves similar performance.