北京2022年8月30日(ri) /美通社/ -- 數(shu)據(ju)已經(jing)成為(wei)(wei)繼土地、勞動力、資本(ben)和技術之后的(de)第(di)五大(da)生產(chan)要(yao)(yao)素，是當代經(jing)濟(ji)社會(hui)發展的(de)基礎資源(yuan)。存儲作為(wei)(wei)數(shu)據(ju)載(zai)體設備發揮著重要(yao)(yao)作用，既要(yao)(yao)滿足當前全球數(shu)據(ju)量高(gao)速增長需求，又要(yao)(yao)保證數(shu)據(ju)存儲安全可靠、讀寫高(gao)效(xiao)精準，從(cong)而為(wei)(wei)數(shu)據(ju)中心提供"穩定(ding)的(de)數(shu)據(ju)存力"。

如(ru)何提(ti)(ti)升數據(ju)存(cun)(cun)儲的(de)可靠性，避(bi)免意外場(chang)景(jing)下的(de)數據(ju)丟失，已成為存(cun)(cun)儲硬件平(ping)臺發展的(de)重大挑戰。浪潮存(cun)(cun)儲從源頭出發，創造性地提(ti)(ti)出了冷熱備電(dian)智(zhi)能切換方案，改進存(cun)(cun)儲系統(tong)對BBU單(dan)元智(zhi)能管理(li)方案，踐行綠色(se)低碳理(li)念，加固備電(dian)質(zhi)量，增強了數據(ju)存(cun)(cun)儲的(de)可靠性。

存儲備電 -- 數據存儲安全的保障

當前業界(jie)存(cun)(cun)儲系統(tong)通常采用(yong)電(dian)(dian)(dian)源PSU（Power Supply Unit）"1+1"冗(rong)余供(gong)電(dian)(dian)(dian)，在供(gong)電(dian)(dian)(dian)之(zhi)外還配置備(bei)用(yong)電(dian)(dian)(dian)池BBU（Battery Back-Up Unit），當機(ji)房市電(dian)(dian)(dian)掉(diao)電(dian)(dian)(dian)，存(cun)(cun)儲系統(tong)實(shi)時(shi)監(jian)測PSU供(gong)電(dian)(dian)(dian)異常，無縫切換(huan)到備(bei)用(yong)電(dian)(dian)(dian)池BBU供(gong)電(dian)(dian)(dian)。BBU提供(gong)持續的供(gong)電(dian)(dian)(dian)能力，確保存(cun)(cun)儲系統(tong)控制(zhi)器寫緩存(cun)(cun)中數據(ju)(ju)，完整而安全的寫入非(fei)易失(shi)性介質，如HDD、SSD等，避(bi)免數據(ju)(ju)丟失(shi)。

為保證數據存(cun)儲(chu)的(de)(de)(de)業(ye)務連續性，機房市電(dian)(dian)(dian)(dian)(dian)意外(wai)掉電(dian)(dian)(dian)(dian)(dian)、市電(dian)(dian)(dian)(dian)(dian)恢復(fu)后能快速(su)恢復(fu)存(cun)儲(chu)系統的(de)(de)(de)業(ye)務，浪潮(chao)存(cun)儲(chu)對備(bei)(bei)電(dian)(dian)(dian)(dian)(dian)設計標準有嚴格要求。比如，浪潮(chao)存(cun)儲(chu)在三(san)年產品生命周期內，備(bei)(bei)用電(dian)(dian)(dian)(dian)(dian)池BBU一次充滿(man)電(dian)(dian)(dian)(dian)(dian)，可滿(man)足(zu)兩(liang)次掉電(dian)(dian)(dian)(dian)(dian)數據備(bei)(bei)份要求；又如，滿(man)足(zu)存(cun)儲(chu)系統高(gao)可靠性要求時，創新地采(cai)用了備(bei)(bei)用電(dian)(dian)(dian)(dian)(dian)池BBU冷熱供電(dian)(dian)(dian)(dian)(dian)切(qie)換策(ce)略，提高(gao)備(bei)(bei)電(dian)(dian)(dian)(dian)(dian)的(de)(de)(de)能效和(he)電(dian)(dian)(dian)(dian)(dian)池的(de)(de)(de)使用壽命，降低(di)BBU電(dian)(dian)(dian)(dian)(dian)池報(bao)廢(fei)的(de)(de)(de)數量，降低(di)對環境的(de)(de)(de)污染。

高端存儲性能提升，存儲備電挑戰升級 

隨著(zhu)數(shu)據量爆發(fa)式增長、存(cun)儲(chu)(chu)(chu)(chu)業務(wu)復雜(za)程度(du)提升，存(cun)儲(chu)(chu)(chu)(chu)硬(ying)件(jian)平(ping)(ping)臺(tai)正(zheng)(zheng)(zheng)朝著(zhu)高密度(du)與高性能(neng)方向(xiang)發(fa)展，傳統(tong)(tong)供(gong)備(bei)(bei)電(dian)(dian)(dian)(dian)(dian)策略難以支撐存(cun)儲(chu)(chu)(chu)(chu)系統(tong)(tong)的(de)穩定性要求。高端存(cun)儲(chu)(chu)(chu)(chu)平(ping)(ping)臺(tai)從系統(tong)(tong)架構到部(bu)件(jian)性能(neng)的(de)升級都伴隨著(zhu)系統(tong)(tong)整體功(gong)(gong)率提升，正(zheng)(zheng)(zheng)常(chang)(chang)(chang)(chang)運(yun)行時存(cun)儲(chu)(chu)(chu)(chu)陣列單(dan)控制(zhi)器功(gong)(gong)率超1700W，掉電(dian)(dian)(dian)(dian)(dian)時刻(ke)控制(zhi)器快速降低功(gong)(gong)耗(hao)(hao)，單(dan)控制(zhi)器功(gong)(gong)耗(hao)(hao)仍超過800W；因(yin)此(ci)單(dan)個BBU的(de)電(dian)(dian)(dian)(dian)(dian)芯(xin)(xin)節(jie)(jie)數(shu)達到12節(jie)(jie)（四(si)串三(san)并），才能(neng)滿(man)足異(yi)常(chang)(chang)(chang)(chang)掉電(dian)(dian)(dian)(dian)(dian)時的(de)備(bei)(bei)電(dian)(dian)(dian)(dian)(dian)功(gong)(gong)耗(hao)(hao)需(xu)求。存(cun)儲(chu)(chu)(chu)(chu)系統(tong)(tong)實時獲取BBU充放電(dian)(dian)(dian)(dian)(dian)次數(shu)與健康狀態(tai)(tai)，BBU單(dan)元檢測(ce)自身狀態(tai)(tai)，如果(guo)發(fa)現(xian)(xian)異(yi)常(chang)(chang)(chang)(chang)可快速定位(wei)(wei)(wei)及修復，延長BBU電(dian)(dian)(dian)(dian)(dian)芯(xin)(xin)的(de)使(shi)用(yong)壽命，保(bao)證BBU供(gong)電(dian)(dian)(dian)(dian)(dian)能(neng)力滿(man)足產(chan)品的(de)需(xu)求。因(yin)為存(cun)儲(chu)(chu)(chu)(chu)系統(tong)(tong)控制(zhi)器功(gong)(gong)耗(hao)(hao)不斷(duan)增大，備(bei)(bei)用(yong)電(dian)(dian)(dian)(dian)(dian)池BBU的(de)電(dian)(dian)(dian)(dian)(dian)芯(xin)(xin)節(jie)(jie)數(shu)不斷(duan)增加，電(dian)(dian)(dian)(dian)(dian)芯(xin)(xin)電(dian)(dian)(dian)(dian)(dian)壓(ya)不平(ping)(ping)衡或損(sun)壞(huai)無法正(zheng)(zheng)(zheng)常(chang)(chang)(chang)(chang)識別(bie)等(deng)質量隱患逐漸突(tu)顯出(chu)來；如果(guo)沒有支持診斷(duan)的(de)BMS（Battery Management System）日志，BBU充電(dian)(dian)(dian)(dian)(dian)異(yi)常(chang)(chang)(chang)(chang)、BBU校驗學習(xi)（評估BBU備(bei)(bei)電(dian)(dian)(dian)(dian)(dian)能(neng)力是否滿(man)足一次備(bei)(bei)電(dian)(dian)(dian)(dian)(dian)需(xu)求）異(yi)常(chang)(chang)(chang)(chang)、BBU供(gong)電(dian)(dian)(dian)(dian)(dian)鏈路偵測(ce)異(yi)常(chang)(chang)(chang)(chang)等(deng)問題則無法準確定位(wei)(wei)(wei)，影(ying)響BBU的(de)故障定位(wei)(wei)(wei)效(xiao)率及使(shi)用(yong)壽命；如果(guo)備(bei)(bei)電(dian)(dian)(dian)(dian)(dian)單(dan)元故障未(wei)及時發(fa)現(xian)(xian)，異(yi)常(chang)(chang)(chang)(chang)掉電(dian)(dian)(dian)(dian)(dian)而BBU不能(neng)正(zheng)(zheng)(zheng)常(chang)(chang)(chang)(chang)供(gong)電(dian)(dian)(dian)(dian)(dian)，嚴(yan)重時會(hui)出(chu)現(xian)(xian)存(cun)儲(chu)(chu)(chu)(chu)系統(tong)(tong)丟(diu)數(shu)據的(de)隱患。

浪潮存儲：智能備電管理技術

浪潮(chao)存(cun)(cun)儲(chu)從(cong)系(xi)統(tong)控制端(duan)(duan)與(yu)備(bei)電(dian)單(dan)(dan)(dan)元(yuan)端(duan)(duan)雙管(guan)(guan)齊下，提出了(le)智(zhi)能(neng)(neng)備(bei)電(dian)管(guan)(guan)理(li)技(ji)(ji)術，將存(cun)(cun)儲(chu)系(xi)統(tong)備(bei)電(dian)單(dan)(dan)(dan)元(yuan)有效電(dian)量(liang)提升(sheng)30%、使用壽命(ming)(ming)延長50%。智(zhi)能(neng)(neng)備(bei)電(dian)管(guan)(guan)理(li)技(ji)(ji)術包含BBU單(dan)(dan)(dan)元(yuan)軟硬一體自(zi)診(zhen)斷方(fang)案(an)、存(cun)(cun)儲(chu)系(xi)統(tong)對BBU的智(zhi)能(neng)(neng)管(guan)(guan)理(li)方(fang)案(an)、冷(leng)熱備(bei)電(dian)智(zhi)能(neng)(neng)切(qie)換(huan)方(fang)案(an)三部(bu)分。通(tong)(tong)過BBU單(dan)(dan)(dan)元(yuan)的監測電(dian)路(lu)與(yu)自(zi)診(zhen)斷算法，實現了(le)BBU狀態監測與(yu)告警(jing)管(guan)(guan)理(li)；通(tong)(tong)過存(cun)(cun)儲(chu)系(xi)統(tong)軟件對BBU單(dan)(dan)(dan)元(yuan)的管(guan)(guan)理(li)，提升(sheng)BBU單(dan)(dan)(dan)元(yuan)故(gu)障定位效率和備(bei)電(dian)系(xi)統(tong)穩定性；通(tong)(tong)過BBU單(dan)(dan)(dan)元(yuan)冷(leng)備(bei)與(yu)熱備(bei)智(zhi)能(neng)(neng)切(qie)換(huan)，BBU單(dan)(dan)(dan)元(yuan)電(dian)池損耗降低(di)30%，提高了(le)電(dian)池的使用壽命(ming)(ming)，降低(di)了(le)電(dian)池報廢數量(liang)和環(huan)境的污染(ran)。

BBU單元軟硬一體自診斷設計方案

浪潮存(cun)(cun)儲(chu)(chu)(chu)通過(guo)自診(zhen)斷算(suan)法為(wei)BBU單元(yuan)提供(gong)精準高效的(de)(de)狀(zhuang)(zhuang)態監測(ce)與異常處理，存(cun)(cun)儲(chu)(chu)(chu)研(yan)發團隊(dui)在設計(ji)前詳細(xi)梳理BMS 軟(ruan)硬件接(jie)口(kou)寄存(cun)(cun)器(qi)、BBU電芯解耦控(kong)制(zhi)參(can)量、狀(zhuang)(zhuang)態保護觸(chu)發閾值等(deng)，用(yong)于監測(ce)狀(zhuang)(zhuang)態的(de)(de)分析(xi)診(zhen)斷。硬件設計(ji)BBU 供(gong)電路(lu)徑(jing)偵測(ce)電路(lu)，實時(shi)監測(ce)BBU供(gong)電路(lu)徑(jing)，跨連接(jie)器(qi)和板卡不同(tong)(tong)位置的(de)(de)電壓(ya)、電流、功率值，作(zuo)為(wei)自診(zhen)斷分析(xi)依據；軟(ruan)件設計(ji)BBU單元(yuan)自診(zhen)斷算(suan)法，BBU單元(yuan)優(you)先查詢電芯物料信息(xi)與當(dang)前狀(zhuang)(zhuang)態進行初診(zhen)斷，初診(zhen)斷無誤后開始對(dui)存(cun)(cun)儲(chu)(chu)(chu)控(kong)制(zhi)信號、充(chong)電信號等(deng)進行實時(shi)記(ji)錄，同(tong)(tong)時(shi)分析(xi)對(dui)外充(chong)放電、對(dui)內校(xiao)驗(yan)學習(xi)等(deng)各種(zhong)狀(zhuang)(zhuang)態下(xia)的(de)(de)參(can)數變化(hua)情況(kuang)。如果狀(zhuang)(zhuang)態參(can)數異常，則(ze)分析(xi)異常原因并進行簡單的(de)(de)自適應調參(can)，同(tong)(tong)時(shi)收(shou)集(ji)異常日志發送給存(cun)(cun)儲(chu)(chu)(chu)系統。通過(guo)流程化(hua)的(de)(de)自診(zhen)斷，可以在存(cun)(cun)儲(chu)(chu)(chu)系統業務(wu)上(shang)線前檢出已(yi)知的(de)(de)大部分問(wen)題(ti)，降低(di)存(cun)(cun)儲(chu)(chu)(chu)系統業務(wu)上(shang)線后BBU單元(yuan)異常的(de)(de)概率。

突破存儲系統對BBU單元智能管理方案

存儲系(xi)統對(dui)BBU單元(yuan)的(de)管(guan)理至(zhi)關重(zhong)要，是備電(dian)流程順利進(jin)行(xing)的(de)核(he)心。存儲系(xi)統基于(yu)BBU單元(yuan)自診(zhen)斷的(de)狀態(tai)信息，從(cong)以(yi)下五方面進(jin)行(xing)智能備電(dian)狀態(tai)監測處(chu)理：

其一(yi)，定期對BBU 供電鏈路偵測(ce)，模(mo)擬存儲系統供電切換流程，提前識別鏈路隱患。

其二，定(ding)期(qi)評估BBU單元儲備的(de)電(dian)量，判定(ding)是否滿(man)足(zu)系統(tong)一次備電(dian)需(xu)求，同(tong)時累計(ji)消除BMS采樣誤差。

其三，存儲系統實時讀取BBU單元電(dian)壓與(yu)電(dian)流、電(dian)芯電(dian)壓及溫(wen)(wen)度(du)、充放電(dian)MOS管(guan)溫(wen)(wen)度(du)，接近BMS內(nei)置閾值(zhi)時報警處(chu)理。

其(qi)四(si)，充(chong)電過(guo)程自動監測存儲(chu)設備功率(lv)，檢測BBU電量是否滿足一(yi)次備電需求，實(shi)時校準(zhun)充(chong)電狀態，同時累計充(chong)放電次數。

最后(hou)，存儲系統對BBU單元BMS狀(zhuang)態寄存器狀(zhuang)態值實時監控，出現異常(chang)后(hou)進入備份供(gong)電(dian)異常(chang)處(chu)理(li)模式(shi)。

上述智能備電(dian)(dian)狀(zhuang)態診斷(duan)方案(an)，將潛在(zai)異常的識別率提升(sheng)了1倍；問題(ti)診斷(duan)完成后，存(cun)儲系統(tong)對(dui)日志(zhi)進行智能分析(xi)，準確(que)定(ding)位出(chu)問題(ti)源頭，例如BBU電(dian)(dian)芯異常、BBU 控制(zhi)模塊異常、存(cun)儲系統(tong)控制(zhi)電(dian)(dian)路(lu)異常、系統(tong)散熱異常等(deng)。

創造性的給出了一種冷熱備電智能切換方案

浪潮(chao)存(cun)儲(chu)系(xi)統(tong)對(dui)(dui)供(gong)(gong)(gong)電(dian)(dian)鏈路定期(qi)(qi)偵測、BBU備(bei)(bei)(bei)電(dian)(dian)能(neng)力定期(qi)(qi)評估，提前(qian)識別供(gong)(gong)(gong)電(dian)(dian)隱患，并基于此進(jin)行(xing)供(gong)(gong)(gong)電(dian)(dian)狀態(tai)(tai)智能(neng)分析，設計了(le)(le)BBU單(dan)元冷(leng)備(bei)(bei)(bei)、熱(re)(re)(re)備(bei)(bei)(bei)智能(neng)切換(huan)方(fang)(fang)案。在1+1冗余，雙(shuang)PSU都正常(chang)狀態(tai)(tai)下采用冷(leng)備(bei)(bei)(bei)以(yi)降(jiang)低(di)備(bei)(bei)(bei)電(dian)(dian)損耗，存(cun)儲(chu)系(xi)統(tong)通過對(dui)(dui)輸(shu)出(chu)電(dian)(dian)壓(ya)電(dian)(dian)流、PWM驅動波(bo)形、溫度采樣值等參數的智能(neng)分析，提前(qian)對(dui)(dui)PSU的工作狀態(tai)(tai)進(jin)行(xing)預測，在單(dan) PSU出(chu)現異常(chang)后，切換(huan)為熱(re)(re)(re)備(bei)(bei)(bei)模(mo)式(shi)，以(yi)保證市電(dian)(dian)異常(chang)時(shi)存(cun)儲(chu)系(xi)統(tong)無縫切換(huan)為BBU供(gong)(gong)(gong)電(dian)(dian)。浪潮(chao)存(cun)儲(chu)打破了(le)(le)傳統(tong)熱(re)(re)(re)備(bei)(bei)(bei)電(dian)(dian)技術對(dui)(dui)BBU壽命(ming)損耗的弊端(duan)，智能(neng)供(gong)(gong)(gong)電(dian)(dian)方(fang)(fang)案中熱(re)(re)(re)備(bei)(bei)(bei)份供(gong)(gong)(gong)電(dian)(dian)時(shi)間占比(bi)不超過10%，BBU單(dan)體待機功(gong)耗由原來的熱(re)(re)(re)備(bei)(bei)(bei)3W.h, 減小(xiao)至(zhi)0.3W.h，一(yi)年(nian)內(nei)充電(dian)(dian)次(ci)數也由450次(ci)減少(shao)至(zhi)50次(ci)左(zuo)右，BBU使用壽命(ming)由不到一(yi)年(nian)延長至(zhi)三(san)年(nian)以(yi)上，使得廢棄BBU對(dui)(dui)環境污染程度大幅度降(jiang)低(di)，貫徹了(le)(le)綠色設計理念。

浪潮存(cun)儲(chu)(chu)秉承"云(yun)存(cun)智(zhi)用 運籌新(xin)數(shu)據"的(de)新(xin)存(cun)儲(chu)(chu)理(li)(li)念，深耕存(cun)儲(chu)(chu)平(ping)臺底(di)層硬件(jian)的(de)創(chuang)新(xin)研發，從源頭做(zuo)起全方位加固(gu)存(cun)儲(chu)(chu)產(chan)品備(bei)(bei)電質量，充分發揮硬件(jian)平(ping)臺的(de)數(shu)據備(bei)(bei)份處理(li)(li)優勢，貫徹落實綠色節能設計理(li)(li)念，打造具備(bei)(bei)極致可(ke)靠(kao)性的(de)高端存(cun)儲(chu)(chu)產(chan)品，保障企業海(hai)量數(shu)據存(cun)得高效(xiao)、存(cun)得可(ke)靠(kao)，護航數(shu)字經濟發展。