modaclouds decision support system for cloud service selection

18
MODAClouds Decision Support System for Cloud Service Selec8on Smra8 Gupta CA Labs, CA Technologies 20 th of March 2015 LDBC Sixth TUC Mee8ng, UPC, Barcelona

Upload: ldbc-the-graph-rdf-benchmark-reference

Post on 10-Aug-2015

45 views

Category:

Technology


0 download

TRANSCRIPT

MODAClouds  Decision  Support  System  for  Cloud  Service  Selec8on  Smra8  Gupta    CA  Labs,  CA  Technologies  

20th  of  March  2015  LDBC  Sixth  TUC  Mee8ng,  UPC,  Barcelona  

2   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

Outline  

Objec8ve  of  the  talk  

Need  for  Decision  Support  System  in  Cloud  service  selec8on  

Overview  of  MODAClouds  DSS  

Key  Features  of  DSS  

Open  Discussions  for  DSS  in  graph  database  community  

3   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

Why  are  we  here?  Decision  Support  System  and  graph  databases  

CALabs  Barcelona  team  has  organically  developed  a  novel  technology  in  the  form  of  Decision  Support  System  as  a  part  of  MODAClouds  project.  

Graph  database  community  is  evolving  and  there  lies  poten8al  to  use  the  DSS  technology  in  addressing  the  graph  database  selec8on  problem  

 Objec8ve  of  this  talk  is  to  start  brainstorming    in  the  community  about  possible  usage  of  the  technology  to  assist  and  enhance  the  use  of  graph  databases  in  enterprises  

4   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

Need  for  Decision  Support  System  in  cloud  service  selec8on  

Mul8ple  dimensions  of  choices  • Trustworthy  Vendors  • Financial,  Legal,  Organiza8onal  and  Technical  constraints  

Mul8-­‐cloud  environment  compa8bility  issues  • Interoperability  • Ease  of  migra8on  • Vendor  lock-­‐in  

Recommenda8on  based  on  different  dimensions  • Cost  • Quality  • Risk  

5   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

What  DSS  does  for  the  users?  

MODA  Clouds  DSS  

Architectural  model  of  deployment  (Tangible  Assets)  

Architectural  deployment  model  enriched  with  user  selected  cloud  services  

MODAClouds User

Cloud  Service  Recommenda8ons  

Technical  and  Business  oriented  Intangible  assets  and  Risk  Acceptability  level  per  asset      

Relevant  Risks  and  Treatments    

Selected  cloud  service  alterna8ves  

6   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

MODAClouds  DSS:  Key  features  

§  Mul8ple  Stakeholder  par8cipa8on  

§  Risk-­‐analysis  based  Requirement  genera8on  

§  Mul8-­‐Cloud  Environment  Compa8bility  

§  Data  gathering  §  Progressive  Learning  

7   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

Mul8ple  actors,  mul8ple  perspec8ves  

§  Different  stakeholders  may  influence  Cloud  Service  selec8on  in  different  ways  

Risk Policy Manager

Decision Owner

Architect

System Operator

Feasibility  Study  

Engineer  

7  

8   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

Asset  defini8on  by  mul8ple  actors    

Business Analyst

Assets  

Product  Innova8on  and  Quality  

Legisla8on  Compliance  

Sales  Rate  Customer  Loyalty  

Market  Awareness  

Business-Oriented Intangible Assets

8  

Technical-Oriented Intangible Assets

Assets  

Data  Privacy  

Data  Integrity  

End  User  Performance  

Maintainability  

Service  Availability  

Cost  stability  

Technical Team

Assets  

Compute  (IaaS)  

File  System  (IaaS)  

Blob  storage  (IaaS)  

Rela8onal  (PaaS)  

Middleware  (PaaS)  

NoSQL  (PaaS)  

Backend  (PaaS)  

Frontend  (PaaS)  

9   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

Risk  analysis  methodology  

Business  Oriented  Intangible  Asset  

Defini8on  

Technical  Oriented  

Intangible  Asset  Defini8on  

Tangible  Assets  Defini8on   Risk  defini8on   Treatments  

Defini8on  

§  Risks  are  iden8fied  on  the  basis  of  protec8ng  the  assets  §  Treatments  are  defined  to  mi8gate  one  or  more  risks  

§  The  outputs  can  be  refined  itera8vely  allowing  users  to  go  back  in  the  methodology  and  update  informa8on  

9  

10   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

Mul8-­‐Cloud  environment    

11   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

Challenges  in  Mul8-­‐Clouds  

11  

• Interoperability:  Risk  of  unexpected  lack  of  replacement  and  consequent  vendor  lock-­‐in  • Migra8on:  Risk  of  non-­‐viable  migra8on  due  to  migra8on  costs  and  complexity  Vendor  lock-­‐in  

• Risk  of  new  security  breaches  due  to  the  increased  complexity  of  the  system  and  new  communica8ons  Security  

• Risk  of  unavailability  of  evidences  in  case  of  fraudulent  ac8ons  Forensic  Evidences  

• Risk  of  costs  unpredictability  Cost  unpredictability  

• Risk  of  lack  of  provider  interest  in  collabora8on  Lack  of  interest  of  CSPs  

• SME  or  companies  using  mul8ple  services  from  mul8ple  vendors  are  unlikely  to  have  the  power  or  the  8me  to  nego8ate.  Increasingly  unstable  cost  and  T&C  problem.  

Lack  of  nego8a8on  on  SLAs  capacity  

12   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

DSS  –  Automa8c  Data  Gathering  Concept  

DSS  Database  

Graph  building  and  data  

transforma8on  

Structured  flat  data  fetch  

JSON  Database  Interface  

XML  

REST  

JSON  

XLSX  

WSDL  

NoSQL  SQL  

Internet  

Flat  files  

Databases  

Graph  

13   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

Progressive  Learning  

Storage  of  User  input  

   

Storage  of    selec8on  of  services  

Storage  of  thresholds  

and  benchmarks  

Subsequent  recommend-­‐a8on  on  selec8on  

Subsequent  recommenda8on  on  services  

•  With  repeated  use  of  DSS,  the  previous  user  logs  and  stored  and  simple  analysis  is  performed  

 •  The  recurring  users  are  recommended  possible  assets  that  might  be  crucial  to  their  firm  

 •  The  users  are  also  recommended  certain  risks  that  have  been  chosen  by  other  users    

•  The  users  are  also  recommended  the  value  of  each  cloud  service  property  based  on  previous  use  of  DSS  

•  With  the  repeated  usage,  DSS  learns  and  improves  its  recommenda8ons  

14   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

Ground-­‐up  developed  Prototype  by  CALabs  

15   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

Open  Source  Technology  Support  for  DSS  

•  hmp://dss.tools.modaclouds.eu/  DSS  open  source  tool  available  at:  

•  hmps://github.com/CA-­‐Labs/DSS  Documented  and  available  in  github  repository  at:  

•  hmp://www.modaclouds.eu/  MODAClouds  Documenta8on  

16   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

Open  Discussion  -­‐  What  are  the  characteris8cs  that  would  define  the  quality  of  a  cloud  graph  database?  -­‐    What  criteria  are  important  in  the  selec8on  of  (cloud)  graph  databases?  

Who  makes  the  decisions  in  industry  to  select  a  par8cular  graph  database  technology  for  a  company?  

How  does  the  graph  database  community  plan  to  manage  legi8mate  customer  concerns  such  as  preven8on  of  vendor  lock-­‐in  and  cloud  outages?  Is  the  synchroniza8on  of  mul8ple  graph  databases  provided  by  different  vendors  possible?  

Is  gathering  data  with  respect  to  different  characteris8cs  that  define  the  quality  of  the  graph  database    an  important  concern?  

How  could  a  DSS  help  for  cloud  graph  database  selec8on?  

17   ©  2015  CA.  ALL  RIGHTS  RESERVED.  

Thank  you  for  your  amen8on!    

Sr.  Research  Engineer  [email protected]  

Dr.  Smra8  Gupta