CAIDA-WIDE Workshop
CAIDA/WIDE Workshop, March 21,
2003
Collaboration between CAIDA and WIDE
-
Monitoring nameserver response time
-
Goals:
- compare monitoring methodologies
- characterize the performance of the current root (or TLD)
nameserver system
-
Currently available:
- passive monitoring of root, gTLD and ccTLD servers
(nevil@caida)
- active monitoring of root and ccTLD servers
({sekiya,kjc}@wide)
- skitter at the root servers (brad@caida)
-
Collaboration items:
- compare measurement data obtained by 3 methods
- how can these methods complement each other?
-
Tentative timeline
-
March:
- run both passive and active monitors on the same machine for
comparison, one in japan (kjc), one in auckland (nevil);
- deploy 3 more passive monitors for gTLD and ccTLD analysis
(nevil);
-
August:
a paper on DNS monitoring methodologies comparing 3 methods
-
Analysis of DNS anomalies
-
Goals:
- study the status of erroneous queries in DNS traffic and find
their causes
- suggest strategies to fix the problems and to protect the DNS
system from the damage
-
Currently available:
- analysis of F-root tcpdumps (wessels@caida)
Paper at PAM2003
- DNS-QC ({kato,kjc}@wide) - WIDE/JPNIC joint effort to report
results of the dns configuration check to the jp domain admins
http://www.nic.ad.jp/ja/dnsqc/index.html (only in
japanese)
- AS112 server log analysis (andre@caida, kato@wide)
Paper at Sigmetrics2003
-
Collaboration items:
- WIDE collects data at the M-root (tcpdump?)
- apply dnstop to japanese ISP DNS traffic
(kato@wide)
- joint CAIDA/WIDE analysis of the data
-
Tentative timeline
-
March:
Internet draft on how to configure name servers with focus on user
side (CAIDA)
-
June:
Internet draft on DNS-QC, common zone configuration errors with
focus on delegation (WIDE)
-
August:
joint report on AS112 analysis (CAIDA/WIDE)
-
Nameserver placement and "best server selection" problem
-
Goals:
- analysis of root/gTLD servers placement
- provide recommendations for future reconfiguration
-
Currently available:
- server selection algorithm modeling (kjc@wide)
- skitter data analysis ({tlee,brad}@caida)
- effects of removing one root server
- analysis of topological similarity
Paper at PAM2003
-
Collaboration items:
- ccTLD server monitoring (nevil@caida,sekiya@wide)
- long-term monitoring of a small set (~10) of ccTLD servers
- combining data obtained by different monitoring methodologies
(cf. item I) for better input to models
({nevil,brad}@caida,{kjc,sekiya}@wide)
- modeling and simulations
- comparison of bind, djbdns, microsoft servers
- possible use of starbed (kjc@wide)? (starbed
consists of 512 PCs, it was built by a WIDE member.
http://www.hokuriku-it.tao.go.jp/english/)
- improving the current model by including timeouts and caching
(wessels@caida,kjc@wide)
- study the impact of non-legitimate queries on the server
placement issues
-
Tentative timeline
-
March:
- select a set of ccTLD servers to monitor (CAIDA/WIDE)
- recommendations on nameserver selection algorithms
(CAIDA/WIDE)
- start simulation in a smaller lab environment (CAIDA/WIDE)
-
New features in the DNS
-
Goal:
- establish methodologies to track the deployment status of the
DNS new features
-
Currently proposed:
- IPv6 deployment, IPv6 root (kato@wide)
- anycast for root servers (kato@wide)
- dnssec (?)
- enum ({sano,fujiwara}@wide)
-
Collaboration items:
- deploy measurement infrastructure for anycast
- interes in tracking IPv6 deployment
Other measurement topics at WIDE
- 10Gb monitoring
- WIDE is about to deploy 10Gb backbone links, possible use of
OC192MON by joerg@nlanr.net (if our budget allows)
- data aggregation/visualization: extension of aguri
({kaizaki,kjc}@wide)
- packet sampling at NSPIXP by Foundry's sFlow
({kato,kjc}@wide)
Meetings schedule
- CAIDA/WIDE workshop in San Diego on March 21st, 2003
- short research presentations by participants to know each
other
- discussion on collaboration items
- PAM2003, April 6-8, 2003, La Jolla
- WIDE paper on active monitoring is accepted
- CAIDA papers on DNS damage and on server placement are
accepted
- Brad will be visiting WIDE during the summer
- 57th IETF, July 13-18, 2003
- 58th IETF, November 9-14, 2003
|
|