Forum: What's the correct way to withdraw a published database/server ?

(forum ? question ?)

Some years ago, one of our public database/server has been published. But now the underlying data are obsolete, the main developers are gone, the PHP code looks like a bowl of spaghetti(*), only a few random users access our data worldwide, nobody wants to work on this....

What would be the correct to withdraw our service ?

My ideas:

add a message in pubmed commons: http://www.ncbi.nlm.nih.gov/pubmedcommons/
add a banner on the web page
if possible, push the code on github/sf/etc...
?

Furthermore, when starting a new service, what would be the correct way to prepare its unavoidable future retirement ?

create a mailing list to alert the users later?
?

(*) that's not my code! that's not my code ! :-)

database practices

• 4,075 views • link

updated 3.4 years ago by Ram

• written 12.3 years ago by Pierre Lindenbaum

I would definitely turn to PubMedCommons now to see if there's any notice. I really liked that feature of the way that was being used.

• 0 views • link

written 12.3 years ago by Mary

1 answer

I think adding a message on the website stating why it is being phased out and what are the plans for phasing it out would be sufficient.

I hate it when I find a great database only to find that the data is 5 years old and most of the functionalities are broken. A simple message on the front page would have saved me the time I spent trying to figure out if the database is any good or not.

• 0 views • link

updated 6.6 years ago by Ram

• written 12.3 years ago by Damian Kao

Moved your comment to an answer ... because now moderation includes this option :-)

I agree with everything and would have only made it a "me too" post.

Sunsetting services is unavoidable, and the only thing that is really irritating with obsolete services is not realizing that they are not maintained anymore and thus waste valuable time on them.

• 0 views • link

written 12.3 years ago by Istvan Albert

I agree that a message is sufficient. If you want to go one step further, how about supplying the dataset or a subset of the data for download to any users who still may want to access it? There are times when 5 year old data are better than no data.

• 0 views • link

updated 6.6 years ago by Ram

• written 12.3 years ago by Katie D'Aco

good point, a simple text dump of the database gzipped up

• 0 views • link

written 12.3 years ago by Istvan Albert

My guess is every Biostar user would agree that transparant life cycle managment (birth > death > archive the data) and release update information should be essential prerequisites for setting up a server or db in the fist place. AWAK this is exasperatingly rare. Sure, if folk are feeling down because funding is pulled/runs out we can understand them not bothering, but there has to be a better way. The PubMed commons announcment as a suplement to the web page announcment is a good idea.

• 0 views • link

updated 6.6 years ago by Ram

• written 12.3 years ago by cdsouthan