Au­to­ma­tion of stor­age re­claim process reaps div­i­dends for Ea­ton

By au­tomat­ing stor­age re­claim process, the com­pany has achieved 90 per­cent re­duc­tion in the num­ber of man-hours re­quired to com­plete the re­claim re­quest, re­sult­ing in sig­nif­i­cant sav­ings

InformationWeek - - Contents - BY VARUN HARAN tweet @varunph

By au­tomat­ing stor­age re­claim process, the com­pany has achieved 90 per­cent re­duc­tion in the num­ber of man-hours re­quired to com­plete the re­claim re­quest, re­sult­ing in sig­nif­i­cant sav­ings

Ea­ton is a di­ver­si­fied power man­age­ment com­pany pro­vid­ing en­ergy-ef­fi­cient so­lu­tions that man­age elec­tri­cal, hy­draulic and me­chan­i­cal power. The com­pany is a global tech­nol­ogy leader in elec­tri­cal prod­ucts, sys­tems and ser­vices for power qual­ity, dis­tri­bu­tion and con­trol, hy­draulics com­po­nents, aero­space fuel, hy­draulic and pneu­matic sys­tems for commercial and mil­i­tary use. Ea­ton has ap­prox­i­mately 100,000 em­ploy­ees and does busi­ness in more than 150 coun­tries.

IT at Ea­ton is pro­vided by Ea­ton In­dia In­for­ma­tion Tech­nol­ogy Cen­ter (EITC), which of­fers a wide va­ri­ety of value-added IT ser­vices to in­ter­nal Ea­ton businesses and func­tional de­part­ments.

Ea­ton’s IT In­fra­struc­ture en­vi­ron­ment sup­ports busi­ness-crit­i­cal ap­pli­ca­tions such as fi­nan­cial and ERP ap­pli­ca­tions span­ning mul­ti­ple busi­ness groups glob­ally. It com­prises of ap­proxi- mately 400 HP-UX and Linux servers that are sup­ported in the back-end with stor­age ar­rays with a to­tal stor­age ca­pac­ity of 1.7 PB. In or­der to use the stor­age space op­ti­mally, there was a recurring need to re­claim de­com­mis­sioned file sys­tems and stor­age space. How­ever, as re­claim­ing stor­age from the servers was a man­ual process, it was prone to er­rors, caus­ing in­con­sis­ten­cies to creep into the disk con­fig­u­ra­tions at the OS level. This led to sta­bil­ity is­sues for the servers over a pe­riod of time. Apart from this, from a process per­spec­tive, the process was time con­sum­ing, leading to greater than ac­cept­able lev­els of MTTR (Mean Time To Re­solve) for disk re­claim re­quests.

In or­der to ad­dress both the chal­lenges, Ea­ton de­cided to ex­plore the pos­si­bil­ity of au­tomat­ing the re­claim process .

MAN­UAL VS. AU­TO­MATED STOR­AGE RE­CLAIM

Ea­ton’s ear­lier re­claim process was a three-step man­ual ac­tiv­ity, per­formed by UNIX & Stor­age re­sources, which re­quired mul­ti­ple hand-offs to per­form their re­spec­tive tasks, as can be seen

from steps 1, 2 & 3 in fig­ure 1. More­over, these tasks could only be per­formed by Level-2 & Level-3 ad­min­is­tra­tors with higher lev­els of skills and spe­cial­iza­tions.

Due to mul­ti­ple hand-offs be­tween re­sources, the aver­age MTTR for com­plet­ing re­claim re­quests was 93 hours. The man­ual process also had a scope for hu­man er­rors that could have caused un­planned out­ages and also caused in­con­sis­ten­cies to creep into the disk con­fig­u­ra­tion of servers. Fi­nally, there were long lead times for un­used space to be avail­able for new projects or or­ganic growth of data­bases and ap­pli­ca­tions.

In or­der to au­to­mate the disk re­claim process, a cross-plat­form tool was de­vel­oped. This tool ex­e­cutes the re­claim process end-to-end, per­form­ing the re­quired tasks at: a) VER­I­TAS Vol­ume Man­ager Level: Re­moves disk ref­er­ences at VER­I­TAS level b) Stor­age ar­ray level: Un­maps and un

masks disks al­lo­cated to the server c) Op­er­at­ing sys­tem level: Re­moves

disk from the OS de­vice tree. All the above tasks are per­formed by in­vok­ing the tool from a sin­gle de­ploy­ment server. Once in­voked, the tool ex­e­cutes the re­quired op­er­a­tions at each of the lev­els men­tioned above. The tool prompts the user for in­puts re­lated to the disks that need to be re­claimed and per­forms the nec­es­sary checks and val­i­da­tions as it pro­gresses through each step/level of the re­claim process.

DE­VEL­OP­ING THE TOOL

This tool was de­vel­oped by us­ing the ex­ist­ing util­i­ties and OS func­tion­al­i­ties avail­able on the HP-UX and Linux servers and in­te­grated us­ing UNIX Shell pro­gram­ming. The tool is de­signed such that the ad­min­is­tra­tor does not need to lo­gin into ev­ery server from which the disks need to be re­claimed. The tool is in­voked from the sin­gle de­ploy­ment server and ac­cesses the tar­get servers through a se­cured shell (ssh) lo­gin.

More­over, the ear­lier man­ual stor­age op­er­a­tions that were per­formed us­ing the SMC GUI are now be­ing per­formed us­ing the sym­cli com­mands, which have been in­cor­po­rated within the tool, thus en­abling the au­to­ma­tion of the stor­age op­er­a­tions. The tool can be used to re­claim disks on HPUX, as well Linux hosts and to re­claim mul­ti­ple de­vices si­mul­ta­ne­ously. De­bug log­ging has been im­ple­mented so that the de­tails of the com­mands be­ing ex­e­cuted can be viewed dur­ing ex­e­cu­tion along with their sta­tus. This out­put is also cap­tured in log files for fu­ture ref­er­ence and trou­bleshoot­ing. Exit codes are cap­tured and checked to val­i­date suc­cess­ful ex­e­cu­tion of a com­mand. The tool also pro­vides an op­tion to de­com­mis­sion file sys­tems hosted on the disks by col­lect­ing user in­puts, dur­ing the course of ex­e­cu­tion.

Most im­por­tantly, this tool was purely de­vel­oped us­ing pre-ex­ist­ing OS util­i­ties and avail­able tools and there­fore no additional costs were in­curred by the or­ga­ni­za­tion dur­ing its de­vel­op­ment.

REAP­ING THE RE­WARDS

The sce­nario post-de­ploy­ment shows a demon­stra­ble im­prove­ment in the ef­fi­ciency of the re­claim process. The process can now be ex­e­cuted by a sin­gle re­source. Ad­di­tion­ally, as the script has in-built er­ror check­ing, val­i­da­tion check-points and is also highly menu driven, this re­claim process can now be as­signed to lesser ex­pe­ri­enced re­sources like Level 1 ad­min­is­tra­tors.

The tool has elim­i­nated the mul­ti­ple hand-offs that were ear­lier re­quired be­tween re­sources and im­proved the over­all sta­bil­ity of the en­vi­ron­ment. The aver­age MTTR for disk re­claim tick­ets is now 10 hours (with each disk tak­ing 2-3 min­utes to be re­claimed), as op­posed to an aver­age MTTR of 93 hours ear­lier due to the de­pen­dency on hu­man in­ter­ven­tion from re­sources.

From an em­ployee per­spec­tive, the so­lu­tion has seen close to 90 per­cent re­duc­tion in the num­ber of man-hours re­quired to com­plete the re­claim re­quest, re­sult­ing in a soft dol­lar sav­ings of ` 7,91,000 per an­num. In ad­di­tion this ser­vice im­prove­ment is help­ing in im­prov­ing the op­er­a­tional ex­cel­lence tar­gets of the Ea­ton’s IT or­ga­ni­za­tion. There is a 50 per­cent re­duc­tion in the num­ber of re­sources re­quired to ex­e­cute the process.

Viswanathan Kr­ish­na­murthi VP – Cor­po­rate IT ( APAC ), Ea­ton Tech­nolo­gies

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.