Data mirrors: Techniques and Issues By Sarah Ponrathnam sarah@iucaa.ernet.in http://www.iucaa.ernet.in/
Overview Introduction… What is Data Mirroring? Techniques of Mirroring… Motives for Mirroring… Mirror site Issues… Popular Mirroring Tools in Unix… Mirroring Experience at IUCAA…
What is data mirroring? Creating an exact duplicate copy in real-time. In terms of web sites, sites are often mirrored to reduce the traffic on one server.
Techniques of Mirroring Replication is one way to solve availability problem. Distributed Servers Cluster Servers Web site Mirrors
Techniques of Mirroring Distributed Servers: Large web destinations such as google, Yahoo have enough capital to set up and to support distributed servers. Eg. www.google.com, www.google.co.in
Techniques of Mirroring Cluster servers:
Techniques of Mirroring Website Mirrors: Mirrors simply compare and pull the contents from a single master web site at a regular intervals and make identical contents available on another computer, ideally closer to the users making use of it.
Motives for Mirroring Load Balancing - yahoo.co.uk yahoo.co.in High Availability - ADS mirrors Multilingual replication - www.debian.org Database Sharing – www.desert.net www.tucksonweekly.com Franchise / Local Versions - quicken.excite.com quicken.com Virtual Hosting - sports.catalogue.com www.accesports.com
Mirroring Issues Maintenance of mirror-sites in different geographical locations. Integrity of the mirrored contents Providing Host-Independent URLS. Initial transfer of data Optimization Economical constraints Providing Efficient routing
Popular Mirroring tools on Unix RSync http://samba.anu.edu.au/rsync/ Wget http://www.gnu.org/ Mirror http://www.wehlus.de/mirror/download.html
Mirroring Experience IUCAA (Inter University Centre For Astronomy and Astrophysics) works closely with ERNET (Educational and Research NETwork) to make this network a content oriented networkhttp://www.iucaa.ernet.in/
ERNET India ? In 1986, then Department of Electronics had initiated a project "ERNET" with the funding from UNDP. The objective was to establish and operate a nationwide Internet for the Indian academic and research community. Now it has become a full pledged ISP known as ERNET India.
ERNET partners
ERNET Network
ERNET NOC at IUCAA
Mirrors available in IUCAA VizieR provides access to the most complete library of published astronomical catalogues and data tables available on line, organized in a self-documented database. http://urania.iucaa.ernet.in
Mirrors available in IUCAA The NASA Astrophysics Data System (ADS) maintains four bibliographic databases containing more than 4 million records. http://ads.iucaa.ernet.in
VOI Data Archives at IUCAA SDSS Sloan Digital Sky Survey – 1 Tb 2MASS 2 Micron All Sky Survey – 194 Gb 2dfGRS 2 degree field Galaxy Redshift Survey – 5.5 Gb 2QZ 2 Degree field QSO Survey – 630 Mb FIRST Survey Faint Images of Radio Sky at Twenty centimeters – 226 Gb
VOI Hardware
Mirror-site under construction Chandra Data Archive (CDA) 424 GB data is available through ftp service. ftp://cdaftp.iucaa.ernet.in Incremental update is being done regularly. Web based CDA service will be provided, once the new release of CXCDS software is made available.
Acknowledgement Dr. Francois Ochsenbein - VizieR Dr. Guenther Eichhorn and Dr.Alberto Accomazzi - ADS Dr. Ramadurai Padmanabhan - CDA
Questions or Comments?