Minimizing Thermal Stress for Data Center Servers through Thermal-Aware Relocation

Algorithm 1

Input:

(i) 𝑆₃ = {All servers at state 𝑆^𝑖₃}

(ii) 𝑆₀ = {All servers at state 𝑆^𝑖₀}

(iii) 𝑆₁ = {All servers at state 𝑆^𝑖₁}

(iv) 𝑆₂ = {All servers at state 𝑆^𝑖₂}

(v) 𝑆^𝑖_TTS= {Thermal profiles and thermal states of all servers}

Algorithm:

(1) For each server Server^𝑖_𝑆3 in set 𝑆₃

(2)  Find a heterogeneous server Server^𝑗_𝑆0 in set 𝑆^𝑖₀ for location switching such that

(3)   𝑆^𝑗_TTS→𝑇^𝑗_maxinlet≥𝑆^𝑖_TTS→𝑇^𝑖_received

(4)   If at least 1 server exists in 𝑆₀

(5)    Switch locations of servers Server^𝑗_𝑆0 With Server^𝑖_𝑆3

(6)   Else If no suitable server exists in 𝑆₀

(7)    Find a heterogeneous server Server^𝑘_𝑆1 in set 𝑆₁ for location switching such that

(8)     𝑆^𝑘_TTS→𝑇^𝑘_maxinlet≥𝑆^𝑖_TTS→𝑇^𝑖_received And

(9)     𝑆^𝑘_TTS→𝑡^𝑘_maxCPU<Server^𝑖_𝑆3→𝑡^𝑖_maxCPU

(10)    If at least 1 suitable server exists in 𝑆₁

(11)     Switch locations of servers Server^𝑖_𝑆1With Server^𝑖_𝑆3

(12)    Else If no suitable server exists in 𝑆₁

(13)     Find a heterogeneous server Server^𝑚_𝑆1 in set 𝑆₂ for location switching such that

(14)      𝑆^𝑚_TTS→𝑇^𝑚_maxinlet≥ 𝑆^𝑖_TTS→𝑇^𝑖_received And

(15)      𝑆^𝑚_TTS→𝑡^𝑚_maxCPU < Server^𝑖_𝑆3→𝑡^𝑖_maxCPU

(16)      𝑆^𝑚_TTS→𝑡^𝑚_minCPU < Server^𝑖_𝑆3→𝑡^𝑖_minCPU

(17)     If at least 1 suitable server exists in 𝑆₂

(18)      Switch locations of servers Server^𝑚_𝑆2 With Server^𝑖_𝑆3

(19)     Else If no suitable server exists in 𝑆₂

(20)      Shift workload from Server^𝑖_𝑆3 to server Server^𝑗_𝑆0

(21) End