Development of Remote Access Systems

11
Development of Remote Access Systems Tanvi Desai LSE Research Laboratory Data Manager Research Laboratory IASSIST 2008: Stanford

description

IASSIST 2008: Stanford. Development of Remote Access Systems. Tanvi Desai LSE Research Laboratory Data Manager. Research Laboratory. IASSIST 2008: Stanford. Cycles. 1990ish inadequate processing power, RAM research PC is a dumb terminal data stored and analysed on central server - PowerPoint PPT Presentation

Transcript of Development of Remote Access Systems

Page 1: Development of Remote Access Systems

Development of Remote Access Systems

Tanvi DesaiLSE Research Laboratory Data Manager

Research Laboratory

IASSIST 2008: Stanford

Page 2: Development of Remote Access Systems

Cycles 1990ish inadequate processing power, RAM

research PC is a dumb terminal data stored and analysed on central server

improved desktop computingdata released to researchers to analyse on PCs

increased dissemination = increased security risk better understanding of confidentiality issues sensitive data no longer released to analyse at

desktop storage on central server

Research Laboratory

IASSIST 2008: Stanford

Page 3: Development of Remote Access Systems

Solutions

Stage 1: complete withdrawal AAAAARGHHH!!!!

Stage 2: Safe Centres inconvenient inequitable often expensive

Stage 3: Remote Solution more convenient - analysis at desktop less expensive, but

Research Laboratory

IASSIST 2008: Stanford

Page 4: Development of Remote Access Systems

Remote Access v Remote Execution

UNECE Glossary on Statistical Disclosure Control Definitions Remote access: On-line access to protected microdataRemote execution: Submitting scripts on-line for execution on disclosive microdata stored within an institute’s protected network. If the results are regarded safe data, they are sent to the submitter of the script. Otherwise, the submitter is informed that the request cannot be acquiesced. Remote execution may either work through submitting scripts for a particular statistical package such as SAS, SPSS or STATA which runs on the remote server or via a tailor made client system which sits on the user’s desk top.

Research Laboratory

IASSIST 2008: Stanford

Page 5: Development of Remote Access Systems

Remote access: examples

LEED project identified remote access systems in Denmark, Netherlands and Sweden

Denmark: remote access using RSASecurIDNetherlands: Remote access using fingerprint

recognition easy to use (though Denmark problems for people

with slower typing speeds and Netherlands problems with recognition)

access at national institutions only qualified staff needed to check output expensive for both users and providers

Research Laboratory

IASSIST 2008: Stanford

Page 6: Development of Remote Access Systems

Remote Execution: examples 1

LEED project identifieed remote executions systems in Czech republic (TREXIMA),

Germany, NetherlandsNetherlands Employment and Earnings Dataset

researchers supplied with synthetic dataset on which to develop statistical programs, which are then sent to to CBS staff to run on full data

Germany LIAB following an initial on-site visit, researchers send

analysis programs to FDZ staff

Research Laboratory

IASSIST 2008: Stanford

Page 7: Development of Remote Access Systems

Remote Execution: examples 2

LISSY (v1 1983)

Data available include Luxembourg Income Study, European Structure of Earnings Survey, Banca d’Italia Survey of Industrial and Service Firms

ESES and Banca d’Italia block some (key?) analyses

cheap system in terms of hardware, setup and support unless depending on level of manual checking

Research Laboratory

IASSIST 2008: Stanford

Page 8: Development of Remote Access Systems

Remote Access

Advantages full access to data providers can maintain very high level of

control

Disadvantages slow return of output access to confidential data experts needed to check output

Research Laboratory

IASSIST 2008: Stanford

Page 9: Development of Remote Access Systems

Remote Execution

Advantages fast job return economical reduced access to sensitive data manual checking not always necessary

Disadvantages sensitive analyses often blocked no access to full dataset not enough research on potentially disclosive

analyses to ensure high level security preparation of synthetic files may be necessary

Research Laboratory

IASSIST 2008: Stanford

Page 10: Development of Remote Access Systems

Conclusions

remotte access/execution can provide equitable access to sensitive data

systems are relatively cheap to run compared with Safe Centres

remote systems fit researchers standard working practices

there is a need to expand remote access beyond national borders primarily in Europe

BUT as the number of remote systems increases can

the qualified staff be found for manual checking?

Research Laboratory

IASSIST 2008: Stanford

Page 11: Development of Remote Access Systems

Thank you

[email protected]

Research Laboratory

IASSIST 2008: Stanford