DataWiki

User Tools

Site Tools

Trace: • ws20140124 • overview
igsn:namespaces

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision 		Previous revision
igsn:namespaces [2012/11/09 08:05]
jklump [Proposal for a hierarchical namespace governance model] changed to IGSN hierarchical namespace governance model 		igsn:namespaces [2016/03/24 10:02] (current)
ulbricht
Line 1: Line 1:
-====== Namespaces ====== +:!: The documentation of the IGSN has moved to [[http://igsn.github.io/]] :!:
- +
-===== IGSN hierarchical namespace governance model ===== +
- +
-==== Recommended Practice ==== +
- +
-  - IGSN e.V. shall adopt a hierarchical model to govern its namespace. +
-  - Top-level namespaces shall be governed and assigned by IGSN e.V. +
-  - [[igsn:glossary#allocating_agent|Allocating Agents]] may apply for the [[igsn:namespace_allocation|allocation of namespaces]] by IGSN e.V. +
-  - Allocating Agents may extend their allocated namespaces into sub-namespaces by adding characters or numbers. The allocated namespaces and corresponding sub-namespaces shall be governed by the respective Allocating Agents. +
-  - IGSN namespaces and IGSN names must conform to the [[igsn:syntax|IGSN syntax]] guidelines. +
- +
-==== Explanation ==== +
- +
- +
-A governance model on the level of IGSN must meet three conditions: +
- +
-  - Flexibility to accommodate the needs of diverse scientific communities +
-  - Assure uniqueness of assigned IGSN names +
-  - Minimal human and technical communication overhead +
- +
-These requirements are best met by hierarchical namespace governance ((Bechtold, S. (2003), Governance in Namespaces, Loy. L.A. L. Rev., 36(3), 1239–1320, [[http://dx.doi.org/10.2139/ssrn.413681]])) +
- +
-=== Flexibility === +
- +
- +
-A hierachical namespace governance model, as is now used by [[http://www.crossref.org/]] and [[http://www.datacite.org]], would satisfy all tree requirements. In fact, DataCite is already a sub-namespace of CrossRef, which in turn is a sub-namespace of [[http://www.handle.net]]. At present, IGSN has been assigned to the namespace 10273 of the Handle namespace. CNRI, the operator of handle.net, has delegated the governance of the sub-namespace 10273 to IGSN e.V. +
- +
-Comment(K Lehnert): IGSNs reported in the literature will have the format IGSN:<namespace> <string>. The namespace '10273' will not  +
- +
-=== Uniqueness and Namespace Exhaustion === +
- +
- +
-In the context of IGSN namespaces it is required to support the 9-character syntax of SESAR and alternative naming conventions used by the major core repositories. At the current state of the discussion the standard prefix for a collection is a string of three characters. This rule allows for potentially 17576 (26^3) namespaces **(!! SESAR also allows numbers in the name space, so it is 36^3 = 46,656 possible namespaces)**. Using the SESAR format of 3+6 characters, the following string of six characters (a-z, 0-9, excluding i and o) allows 1.55 x 10^9 samples per namespace, which should be sufficient for most purposes. +
- +
-To assure unique names for long term operation of the system and to accommodate the requests from the core repositories a few points should be considered: +
- +
-  - Many collections are much smaller than 1,500 million samples. In most cases four characters would be sufficient, even if the name suffixes are only numbers. +
-  - Using longer collection prefixes as sub-namespaces would multiply the number of collection namespaces by orders of magnitude while still providing a sufficient number of names for small collections of about 10,000 samples. +
- +
-The proposed hierarchical namespace governance model would be similar in structure to the IP address space (IPv4) or to telephone numbers in Europe. +
- +
-The image below shows the geographical distribution of the first digit of telephone area codes in Germany. Large cities with many subscriber lines have short area codes (e.g. Berlin = 030, Munich = 089, Hamburg = 040), while smaller cities have longer area codes (Potsdam = 0331, Freiburg = 0761), whereas small towns have long area codes (e.g. Groß Pankow = 033983). In turn, the number of digits of a local subscriber line is long (up to eight digits) in large cities, but may be much shorter in small towns.   +
- +
-{{:igsn:area_codes_germany.png?left&200|Geographical distribution of telephone area codes in Germany}}  +
- +
-Source [[http://de.wikipedia.org/wiki/Telefonvorwahl]] +
- +
-=== Communication Overhead === +
- +
-In a hierarchical namespace governance model the Allocating Agent does not need to negotiate the allocation of sub-namespaces with IGSN e.V. but may solely negotiate with its clients. This is analogous to the current practice in assigning DOI names, or handle names. By delegating parts of the namespace governance to Allocating Agents the communication overhead between Allocating Agents and IGSN e.V. will be minimised. +
- +
-==== Mnemonic Names ==== +
- +
-Principal investigators applying to an Assigning Agent for a namespace might prefer mnemonic names. However, it is foreseeable that many mnemonic namespaces will soon be allocated. When this point is reached only non-mnemonic namespaces can be allocated. This situation is similar to "vanity plates" in car licence plates. For example, in Berlin all car licences in the namespace B-MW nnnn have already been allocated, this particular namespace has been exhausted. An Assigning Agent should take care to reserve "valuable" namespaces for their major clients (e.g. ODP, IODP, ICDP, ...). +
- +
-==== Legacy Namespaces ==== +
- +
-The current number of namespaces already assigned by SESAR is small enough to leave already assigned namespaces (legacy namespaces) in place. +
- +
-======  ====== +
- +
-[[igsn:start|Back]] to overview+
igsn/namespaces.1352448304.txt.gz · Last modified: 2012/11/09 08:05 by jklump

Page Tools

Donate Powered by PHP Valid HTML5 Valid CSS Driven by DokuWiki