Construction of an Integrated Management Platform to Support DistributedComputing Systems

(extended abstract of the MSc dissertation)

Sara Filipa Coutinho Barbas ValenteDepartamento de Engenharia Informatica

Instituto Superior Tecnico

Advisor: Professor Jose Borbinha

Abstract—Currently managing an enterprise or institutionwould be impossible without the use of Information Technology.Providing quality services that allow the user an easy and trans-parent handling of information is a daunting task for anyonewho provides it. ITIL is a framework that can help to providesolutions that improve practice management, integration andmanipulation of information. ITIL describes ”WHAT TO DO”and not ”HOW TO DO”, explaining in detail all the processesbut does not indicate how they should be implemented.

In this document we proposed to implement Change Man-agement and Configuration Management processes consideredto be a priority to support the daily management operationsof an organization.

Using Action Research Method, this thesis was applied toa scientific and technical association of public utility whoseresearch, development and implementation is also focused inthe field of managing a distributed computing infrastructure.

A prototype system was built, and now is possible to makefaster changes on the hardware infrastructure. There is alsothe guarantee that, by the end of the change, this informationis recorded in the CMDB correctly and updated.

I. INTRODUCTION

Currently, managing an enterprise or institution wouldbe impossible without the use of Information Technology.Providing quality services that allow the user an easy andtransparent handling of information is a daunting task foranyone who provides it. ITIL is a framework that canhelp to provide solutions that improve practice management,integration and manipulation of information. ITIL describes”WHAT TO DO” and not ”HOW TO DO”, explaining indetail all the processes but does not indicate how they shouldbe implemented.

In this document we proposed to implement ChangeManagement and Configuration Management processes con-sidered to be a priority to support the daily managementoperations of an organization.

Using Action Research method, this thesis was applied toa scientific and technical association of public utility whoseresearch, development and implementation is also focused inthe field of managing a distributed computing infrastructure.

A prototype system was built, and now is possible to makefaster changes on the hardware infrastructure. There is alsothe guarantee that, by the end of the change, this informationis recorded in the CMDB correctly and updated.

A. Motivation

Nowadays the Internet is used on a global scale and isthe main veicule to process information. Modern scienceproblems require large computation time to be solve andgenerate a high volume of data. To face this challengethe scientific community and enterprises use solutions ofdistributed computing such as the Grid [1] [2] and Cloud [3]Computing.

With the adoption of distributed computing, Data Centersbecome harder to manage: they are responsible for managingtheir local infrastructure but they are also part of broadernetworks that can be private, national or even global. Inthis context, Data Centers are responsible for managing theequipment (physical and virtual) and services that can beused by a large communities of users, and should be con-stantly monitored to ensure the quality of services requiredby clients of joint infrastructure.

In sum, to administer a modern Data Center is necessaryto handle all its components: the Physical and networkinfrastructure, the services provided and also understand therelations and dependencies among them.

In order to help to solve this complexity there is theITIL [4] [5] [6] [7] [8] [9] ”a set of best-practices forInformation Technology (IT) service management that focuson aligning IT services with the need of the business.”

ITIL is a vast compilation, with several areas, but the ITILService Support area is the practice of disciplines that enableIT services to be provided. Service Support deals with theprocesses needed to maintain daily operations, and for thisreason should be the first to be implemented.

ITIL service Support include two processes inextricablylinked: Change Management and Configuration Manage-ment. The CMDB (Configuration Management Database)is a repository of data related to all components of theinformation system and a central part of ConfigurationManagement process.

B. Research Questions

The goal of this thesis is to understand the processesinvolved when managing the physical and network infras-tructures of a Data Center: what are the processes involved,

1

which are considered do be a priority, who are the personsinvolved in the processes, and their responsibilities.

Is also part of the scope of this thesis to understand whattools are on the market that can automate and implementthe processes found.

In sum the goal is to find an answer to the followingquestions:

1) How to create the Change Management process toallow, in a coherent and common manner, to managechanges in the physical and network infrastructures?

2) How to create the Configuration Management processand the CMDB, so that the information about theinfrastructure, the relations between its various itemsare known, are globally accessible and updated?

3) What tools can be used and how they can be adaptedto implement Change and Configuration processes?

C. Research Method

The research method used in developing the thesis is Ac-tion Research [10]. This method indicates that the researchershould try a theory with individuals in real environments.The organization where this thesis applies the theory isthe Data Center of Laboratorio de Instrumentacao e Fısicaexperimental de Partıculas (LIP) in Lisbon. The context ofthis institution is explained in Section III.

II. RELATED WORK

The related work of the thesis consists of studies com-paring tools for managing and operating the network andservices infrastructures of a Data Center.

Several open source solutions for network monitoring areanalyzed: Nagios, OpenNMS, and Zenoss [12] and Hyperic,Lithium e o Zabbix [13].

Next, ITIL (Information Technology Infrastructure Li-brary) [16] [4] and the five ITIL Service Support pro-cesses [4] [5] [6] [7] [8] [9] are introduced and explained,giving special attention to Change Management and Config-uration Management processes.

This Section ends with a study [26] that compares soft-ware products according to the capacity they have to imple-ment Change Management and Configuration Managementprocesses.

III. MY WORK

In this section the LIP institution is presented, and themain problems in the procedures undertaken to managethe Data Center of the institution are identified. Given theassessment made to LIP Data Center a set of Functional,Non Functional and Business requirements were proposed.In order to accomplish these requirements two systems weredesigned (according to ITIL): Change Management Systemand Configuration Management System. The section endswith the description of the technologies involved, (RequestTracker and Zenoss) and how they are adapted, to implementthe proposed systems.

A. LIP ContextLIP is a scientific and technical association of public util-

ity that has for goal the research in the fields of ExperimentalHigh Energy Physics and Associated Instrumentation. Themain research activities of the lab are developed in theframework of large collaborations at CERN and at otherinternational organizations and large facilities in Europe andelsewhere, such as ESA and NASA.

In Lisbon and Coimbra there are two Data Centers hostingthe IT infrastructure of the institution. The computationteam is also responsible for coordinating the operation ofthe Iniciativa Nacional Grid (INGRID) and for managingthe computational systems of the Grid Central Node incollaboration with FCCN.

B. Main Problems in Managing LIP Data CentersA large number of interviews to the computation team

were made in order to assess how the daily operations weremade. The main conclusions were:

• the documentation of how processes are made is not upto date

• the database that holds the IT infrastructure is obsolete• each functionary holds in his private area the documents

and procedures needed for doing his tasks• every intervention is performed in a ad hoc basis

without properly evaluating the impact results• The flow of information needed when tasks are per-

formed for several people is not being recorded

C. Functional, Non Functional and Business RequirementsThe main business requirement is that the systems to

be designed should be implemented using open sourcetechnologies, and capable of integrating technology alreadyexisting in the institution. The main functional requirementis that the system to be implemented should allow theplanning of interventions (adding, replacing, remove) in theIT infrastructure.

D. Change Management SystemPlanning an intervention in the IT infrastructure is plan-

ning a change to be made, so a Process for Change Man-agement composed by several activities was designed. (seeFigure 1) The system functionalities were established andalso the use cases of the system. (See Figure 2)

E. Configuration Management SystemWhen a change is made the IT infrastructure changes. One

or several configuration items can be altered. The CMDBis the repository that holds these items and is a part ofConfiguration Management process.

To achieve this, the Configuration Management Processfor LIP was designed. The activities that composed theprocess are: Planning, Identification, Specification, Controland Auditing and Verification.

In the Planning phase, the configuration items and theCMDB were identified.

The system functionalities were established and also theuse cases of the system. (See Figure 4)

2

F. Request Tracker to Implement Change ManagementRequest Tracker (RT) [29] is a tracking system which can

be used for bug tracking, help desk ticketing, customer ser-vice, workflow processes, change management and networkoperations. In the context of this thesis is used to implementchange management system and managing the workflow.

It has the following qualities: accessibility, ease of use,multiuser, ability to track history, immutable history, flexibleviews, access control, dependency management, notifica-tions and customizable workflow.

In RT each request or piece of work is called a Ticket.This ticketing system allows to register an event or a ticket,assign an owner, or person responsible to the ticket, assignadditional interested parties to the ticket, track changes tothe ticket, launch activities based on ticket status and priorityamong others.

RT Configuration: RT was configured according to theproposed change management process. The main conventionis based in the correspondence between a Request forChange (RfC) and a ticket in a queue that aims changes inthe hardware infrastructure; so all changes will be activitiesrelated to hardware attributes devices. The ticket structurewas augmented with extra custom fields in order to supportthe attributes of a RfC and also to support the device nameand type which will be changed. Table I summarizes thecustom fields in a ticket RT:

Formalismo RfC Ticket RT Preenchimento

RfC Number Ticket Id AutomaticCreation Date Created AutomaticExpected Start Date Starts AutomaticStart date Started ManualExpected End Date Due AutomaticDescription Subject Manual/MandatoryPriority Priority AutomaticUrgency Urgency Manual/MandatoryCategory Category level1 Manual/MandatoryCategory Category level2 Manual/MandatoryChange Type Change Type Manual/MandatoryDevice Name Device Name Manual/MandatoryDevice Name Device Name Manual/Mandatory

Table ICORRESPONDENCE BETWEEN A TICKET AND A RFC.

Workflows in RT are implemented using a mechanism ofscrips, allowing the definition of a work hierarchy.

When a change is closed, a configured action take placein RT, and the description of the hardware device and hisattributes is written in a XML file.

G. Zenoss to Implement Configuration ManagementZenoss [34] is a open source system for managing servers

and networks. Zenoss monitors the IT infrastructure, includ-ing the network, the servers, air conditioning, heating sys-tems, and it also monitors performance and the availability. Ithas a CMDB of the IT hardware and software infrastructure.

Larry Klosteboer [33] indicates two approaches that canbe used to introduce the devices in the CMDB: the firstapproach (waterfall) is to gather as much of data as possiblefrom throughout the environment, and then integrate andlink data, to assemble a complete configuration managementpicture. This is a more direct approach, but results in aoverwhelming amount of data, which delays the completionof the task. The second approach (trickle) is to create integra-tion points in each of the key areas of operational processesthat are likely to deal with configuration data. By executingnormal operations with these integration points, data relatedto configuration items trickles into the CMDB as those itemschange. This is less risky approach , but delays the benefitsof having a complete image of Configuration Managementprocess. The second approach is the one adopted in thisthesis: a change in a IC under Configuration Managementprocess, always arises as a result of a change request certifiedand approved. Subsequently the information related to theIC is introduced in Zenoss CMDB.

1) Implementation of Activities of GC Process in Zenoss:

Identification, Specification and Control: Zenoss providesa suitable domain model for the LIP institution. If in thecourse of activities, this model needs to be extended, Zenossprovides methods (Zenpacks) that allow the creation of newdevice classes and new methods to monitor them.

So, the Configuration Manager must take in considerationall the activities proposed for Configuration Managementprocess and carefully assess what steps should be undertakenin each activity and if there is the need to extend the model.

The Configuration manager should only trigger the Con-trol Activity after ensuring (covering other activities pro-posed) that there are no inconsistencies between the infor-mation to be added/modifieded/removeed and the contentsof the CMDB.

It is possible to introduce information in Zenoss in threedifferent ways:

1) through the graphical interface2) through the interactive interpreter running on the com-

mand line, called Zendmd3) through a dedicated APIIn order to automate the process of inserting contents in

Zenoss CMDB, a python application which uses the featuresprovided by Zenoss API was implemented. This applicationis able to insert XML content provided by RT via http,allowing the Configuration Manager to automatically byexecuting the application in the command line, to establishconnections to Zenoss CMDB and insert new devices.

Auditing and Verification: The tool itself can trigger theAuditing activity taking into account the capabilities todiscover and monitor devices. Zenoss is able to discover var-ious types of devices, scan and understand configuration in-formation and provides modeling capabilities, and is exten-sible.

The Zenoss provides several ways to find and introducedevices and their components: manual discovery and auto-matic discovery .

3

IV. EVALUATION

This section discusses the results achieved by assessingwhether the research questions presented in Section I-B areanswered.

1) Answer to Question 1:A Change Management process, according to ITILframework, was proposed: the activities were es-tablished, roles were created, defining the personsresponsible for executing them, and changes werecategorized. The main features of a system, that arealigned with the common practices and key proceduresof LIP institution and that implements the proposedChange Management were identified.

2) Answer to Question 2:A Configuration Management process, according toITIL framework, was proposed: the activities wereestablished, roles were created, defining the personsresponsible for executing them. The assets of theinstitution that must be maintained in the CMDB, andthe flow of implementation of CMDB were defined.

3) Answer to Question 3:The Change Management system was achieved usingRequest Tracker tool: its large capacity of adjustmentto heterogeneous environments, the flexibility in con-figure workflows and the simplicity of configurationallowed the implementation of Change Managementaligned with ITIL framework.The Configuration Management system was achievedusing Zenoss tool: Zenoss domain model proved tobe appropriate to implemented in the LIP institution.The features provided, such as monitoring and auditingproved to be an asset in the implementation process.

V. CONCLUSIONS

The main questions that this thesis aims to answer havebeen solved. There is a prototype system in operation andknowledge in LIP institution that enables to continue futuredevelopments of the system.

It is concluded that the Change Management and Con-figuration Management systems were designed and imple-mented: it is now possible to make changes on the hardwareinfrastructure, where each participant in the workflow knowswhat is the status of the task. There is the guarantee that atthe end of the change, this information is recorded in theCMDB and updated correctly.

VI. FIGURES

Figure 1. Proposed Change Management Process.

Figure 2. Use Cases for Change Management System.

4

Figure 3. Activities of Configuration Management Process proposed toLIP.

Figure 4. Use Cases for Configuration Management System.

Figure 5. Ticket Template configured to Perform a Change.

5

REFERENCES

[1] Judith M. Myerson. Cloud Computing Versus Grid Computing.Service Types, Similarities and Differences, and Things toConsider, 3 March 2009,

[2] Foster Ian, Kesselman Carl. The Grid2: Blueprint for a NewComputing Infrastructure. Elsevier Inc, 2004

[3] Chappell David. A Short Introduction to Cloud Platforms: AnEnterprise-Oriented View. Chappell and Associates. 2008.

[4] Office of Government Commerce. Itil Service Support. TheStationery Office. 2000.

[5] Office of Government Commerce. Itil Service Delivery. TheStationery Office. 2000.

[6] Office of Government Commerce. Planning To ImplementService Management. The Stationery Office. 2000.

[7] Office of Government Commerce. Application Management.The Stationery Office. 2000.

[8] Office of Government Commerce. ICT Infrastructure Manage-ment. The Stationery Office. 2000.

[9] Office of Government Commerce. Security Management. TheStationery Office. 2000.

[10] Baskerville, Richard L. 1999. Investigating Information Sys-tems with Action Research. Commun. AIS, 2, 4.

[11] Office of Government Commerce. Best Practice for ServiceSupport, Managing IT Services. The Stationary Office. 2000

[12] Jane Curry. Open Source Management Options. White Paper,Skills 1st Ltd. September 30th, 2008.

[13] Scott Stone. Monitoring Systems Comparison White Paper.The Forbin Group. June 14, 2007.

[14] Will Capelli. Magic Quadrant for Application PerformanceMonitoring White Paper. Gartner RAS Core Research Note,G00173116. 18 February 2010. RA2 02222 011.

[15] David Williams, Debra Curtis. Magic Quadrant for ITEvent Correlation and Analysis White Paper. Gartner RASCore Research Note, G00208774. 13 December 2010. RV3A412152011.

[16] Long John: ITIL Version 3 at a Glance. Springer.

[17] COBIT student Book. IT Governance institute.

[18] Framework, Management Guidelines, Information Systems:Audit, Control Foundation and Objectives. COBIT SteeringCommittee and the IT Governance Institute. IT GovernanceInstitute . COBIT R©, 2000, E.U.A.

[19] Mary Beth Chrissis, Mike Konrad, Sandra Shrum. CMMI forDevelopment: Guidelines for Process Integration and ProductImprovement (3rd Edition). Addison-Wesley Professional, 20March 2011

[20] ITIL course. Electronic Data Systems. 2004.

[21] ITIL Configuration Management Requirement Analysis andPrototype Implementation thesis. Jurgen Langthaler. Universi-dade de Tecnologia de Viena 2007

[22] Greiner, Lynn. ITIL: The International Repository of ITWisdom. netWorker, 11(4), 9-11. 2007.

[23] Van Bon. Foundations of IT Service Management based onITIL V3. Zaltbommel: Van Haren Publishing. January 2007

[24] Mattila, Antti. Best Practices for Network Infrastructure Man-agement - a Case Study of Information Technology Infras-tructure Library (ITIL). M.Phil. thesis. Helsinki University ofTechnology. 2008.

[25] Spremic, M., Zmirak, Z., Kraljevic, K. IT and BusinessProcess Performance Management: Case Study of ITIL Im-plementation in Finance Service Industry. Pages 243-250 of:30th International Conference on Information Technology In-terfaces. June 2008.

[26] ITIL: Microsoft and Open Source, White Paper. MicrosoftCorporation, October 26, 2007.

[27] Filipe Crespo Martins. Implementing ITIL Change Manage-ment thesis. Instituto Superior Tecnico, Universidade Tecnicade Lisboa. 2010

[28] Anthony L.A.Baron, Woodinville, WA (US); Nigel G. Cain,Redmond, WA, (US). CMDB SCHEMA. United States PatentApplication Publication. Pub. No.:US 2006/0004875 A1. Pub.Date: Jan.5, 2006.

[29] Jesse Vincent, Robert Spier, Dave Rolsky, Darren Chamber-lain, Richard Foley. RT Essencials. O‘REILLY, 2005

[30] Jane Curry. Zenoss Discovery and Classification White Paper.Skill 1st Ltd. February 2009

[31] Methods of monitoring processes with Zenoss White Paper,Jane Curry. Skill 1st Ltd. February 2009

[32] Zenoss Enterprise Architecture Overview, White Paper. Copy-right 2010 Zenoss, Inc.

[33] Klosterboer, Larry. Implementing ITIL Configuration Man-agement. IBM Press, 2008

[34] Zenoss Developer’s Guide Version 3.0. Zenoss, Inc.

[35] http://www.gnu.org/copyleft/gpl.html

[36] http://www.lighttpd.net/

[37] http://perl.apache.org/start/index.html

[38] http://www.fastcgi.com/devkit/doc/fcgi-spec.html

[39] http://requesttracker.wikia.com/wiki/CustomizingWithCallbacks

[40] www.zope.org

6