You are here: Home > Publications > RIPE Labs > Denis Walker > Update Statistics for the RIPE Database

Update Statistics for the RIPE Database

Denis Walker — 22 Feb 2012
In this article we provide some analysis on the updates received by the RIPE Database during 2011 and in the first two months of 2012.

Introduction

We were asked to look at the number of updates received for RIPE Database objects and analyse how and where they come from. We separated the incoming updates into three categories:

  • Email updates
  • Syncupdates
  • Other (internal processes)

For the purpose of this analysis we do not differentiate between Webupdates, Syncupdates or the RESTful API. At the back end, these all use the Syncupdates process and are grouped together in that category.

To determine where the updates were coming from, we chose to identify them in relation to MNTNER objects rather than by IP or email address. For each update message, we identified which MNTNER object(s) authorised the updates of the objects within that message. These were then grouped by the number of updates authorised by each MNTNER object found in each of the two categories.

For ease of reading we summarised the number of updates into ranges. The raw data is added at the end.

Two groups of data are provided covering 1 January - 20 February 2012 and the whole of 2011.

Summaries of data analysed

2012 RIPE Database updates (up to 20 February)

In the first section we show update statistics collected between 1 January and 20 February 2012.

In Table 1, you can see the number of updates received sorted by update method and update type. The total number of objects submitted between 1 January - 20 February 2012 was 2,675,721.

Update methods:       Update types:  
sync 600613     RPSL object 697888
mail 111496     help 25
other 251     spam 14447

Table 1: Update methods and update types in 2012

Update methods 2012

Figure 1: Update methods used between 1 January 2012 - 20 February 2012

 

Sync updates MNTNER count 2012

Figure 2: Number of MNTNERs doing a set number of Sync updates in 2012

In Figure 2 the number of MNTNER objects are grouped by number of updates. For example, 79% of the MNTNER objects that authorised changes in sync update messages submitted between 1 and 10 sync updates to the RIPE Database in this period. Only 2% of the MNTNER objects submitted between 101 and 1,000 sync updates.

Mail updates MNTNER count 2012

Figure 3: Number of MNTNERs doing a set number of Mail updates in 2012

In Figure 3 the number of MNTNER objects are grouped by number of updates. For example, 73% of the MNTNER objects that authorised changes in mail update messages submitted between 1 and 10 mail updates to the RIPE Database in this period. Only 1% of the MNTNER objects submitted between 1,001 and 10,000 mail updates.

Tables 2 and 3 show the actual numbers for how many (ranges of) updates are done by how many maintainers. The first one shows sync updates, and the second one shows mail updates.

Nr. of Sync Updates -->done by--> Nr. of MNTNERs
1-10   4821
11-100   1172
101-1000   104
1001-10000   10
10000+   6

Table 2: Number of Sync updates done by number of maintainers

Nr. of Mail Updates -->done by--> Nr. of MNTNERs
1-10   1252
11-100   373
101-1000   74
1001-10000   15
10000+   2

Table 3: Number of Mail updates done by number of maintainers

 

2011 RIPE Database updates

In the next section we show update statistics collected between 1 January and 31 December 2011. In Table 4, you can see the number of updates received sorted by update method and update type. The total number of objects submitted in 2011 was 5,033,598

Update methods:       Update types:  
sync 1389022     RPSL object 3083190
mail 1834380     help 30639
other 11226     spam 120799

Table 4: Updates methods and update types seen in 2011

Update methods 2011

Figure 4: Update methods used in 2011

Sync updates MNTNER count 2011

Figure 5: Number of MNTNERs doing a set number of Sync updates in 2011

In Figure 5 the number of MNTNER objects are grouped by number of updates. For example, 71% of the MNTNER objects that authorised changes in sync update messages submitted between 1 and 10 sync updates to the RIPE Database in this period. Only 3% of the MNTNER objects submitted between 101 and 1000 sync updates.

Mail updates MNTNER count 2011

Figure 6: Number of MNTNERs doing a set number of Mail updates in 2011

 

In Figure 6 the number of MNTNER objects are grouped by number of updates. For example, 74% of the MNTNER objects that authorised changes in mail update messages submitted between 1 and 10 mail updates to the RIPE Database in this period. Only 19% of the MNTNER objects submitted between 11 and 100 mail updates.

Tables 5 and 6 show the actual numbers for how many (ranges of) updates are done by how many maintainers. The first one shows sync updates, and the second one shows mail updates.

Nr. of Sync Updates -->done by--> Nr. of MNTNERs
1-10   15333
11-100   5634
101-1000   658
1001-10000   62
10000+   10

Table 5: Number of sync updates done by number of maintainers in 2011

 

Nr. of Mail Updates -->done by--> Nr. of MNTNERs
1-10   4235
11-100   1085
101-1000   347
1001-10000   56
10000+   19

Table 6: Number of Mail updates done by number of maintainers in 2011

Raw data

The raw data for 2011 and 2012 is available on request. This data expands on Tables 2, 3, 5 and 6. So for example in Table 6 we see that 4,235 MNTNERs submitted 1-10 mail updates. The raw data includes the actual numbers, for example 1,680 MNTNERs submitted 1 update, 972 MNTNERs submitted 2 updates, etc. If you want this data please contact ripe-dbm _at_ ripe _dot_ net .

 

2 Comments

Daniel Karrenberg says:
23 Feb, 2012 10:36 AM
Do I read this data correctly when I conclude that there has been a shift from mail towards the other update methods from 2011 to 2012? If so was that trend already visible in 2011? What do you think has caused this trend?

I am also interested to know why there are a few maintainers doing so many updates. Are these what we used to call "NULL" updates, e.g. not changing anything in most updates they touch? And if they are changing anything for so many objects, what is it in general terms?
Denis Walker says:
23 Feb, 2012 12:23 PM
Dear Daniel

Maybe one and a half months into 2012 is not enough of a data set to make too many judgments on annual stats. But looking at some monthly figures there has been a gradual shift from mail updates to sync updates. Having re-written the Webupdates and Syncupdates services and provided a RESTful API has certainly made these update methods more reliable and easy to use.

Regarding your second question, since the beginning of this year one organisation has been submitting a series of updates on a daily basis. Each of these updates contains hundreds of INET6NUM objects that have not changed. So the count of 'No Operation' results from the update service massively increased in 2012.

As a side note, we generated these particular stats in response to a specific question from the community to assist in the decision making process about the use of passwords in email updates.
Add comment

You can add a comment by filling out the form below. Comments are moderated so they won't appear immediately. If you have a RIPE NCC Access account, we would like you to log in.