Print

Print


All -

I have been watching the debate with interest.  Let me add my two cents worth:

1. multiple copies is a sensible approach as long as they are connected by (a) a coherent data management policy; (b) mirroring in case of updates (or redirection to new versions);

2. the location of the copies should depend on proximity of access - moving data is expensive and has high latency especially with the ever-increasing size of datasets.  This might argue for a local institutional copy (for the research team that generated the dataset) and 2 or more 'distant' locations.  If the dataset is part of an international research collaboration there are arguments  in favour of copies in different zones off the world (e.g. Americas, Europe, Far-East like the W3C model) to take advantage of 'quiet time' which becomes more relevant in (3) below;

3. ideally one would not shift datasets but access them.  That is the dataset host should provide a service for selection of records to reduce data transmission. This I done where research projects have portals - but what is their sustainability?  In the best case the host should allow mobile code to execute on the host to perform locally the selection and data processing - but this is fraught with dangers in terms of capacity, performance, security, privacy, trust.....   However among trusted data centres this may be possible; I doubt institutions (universities) would be willing.

4. each of the above leads to interesting questions of costs and payments and the evergreen problems of rights/licences and their management;

5. all need to have provenance information and metadata covering discovery, context and 'schema level' to connect software to the dataset

Just another angle on the discussion

Keith

-------------------------------------------------------------------------------------------------------------------
Keith G Jeffery Consultants
Prof Keith G Jeffery
E: [log in to unmask]
T: +44 7768 446088
S: keithgjeffery

Past President ERCIM www.ercim.eu   ([log in to unmask]) 
Past President euroCRIS www.eurocris.org   
Past Vice President VLDB www.vldb.org 
Fellow (CITP, CEng) BCS www.bcs.org 
Co-chair RDA MSDWG https://www.rd-alliance.org/ 
Co-chair RDA MIG https://www.rd-alliance.org/ 
C0-chair RDA DICIG https://www.rd-alliance.org/
----------------------------------------------------------------------------------------------------------------------------------
The contents of this email are sent in confidence for the use of the 
intended recipient only.  If you are not one of the intended 
recipients do not take action on it or show it to anyone else, but 
return this email to the sender and delete your copy of it.
----------------------------------------------------------------------------------------------------------------------------------


-----Original Message-----
From: Research Data Management discussion list [mailto:[log in to unmask]] On Behalf Of John Milner
Sent: 03 April 2014 09:28
To: [log in to unmask]
Subject: Re: The Value & Impact of Data Sharing & Curation

I don't really think that's a very sensible approach. The key issue to ensure that the data is safe. The LOCKSS model suggests multiple copies in different places and that's sensible (though there is a gotcha, see below). However it may be that the third party's back up and security arrangements fulfil that need. It is certainly the responsibility of the data owner and his/her institution to check that arrangements are satisfactory. So keeping one copy locally and one at the repository only works if the preservation strategies are compatible (Gotcha!)

John

-----Original Message-----
From: Research Data Management discussion list [mailto:[log in to unmask]] On Behalf Of Richard Rankin
Sent: 03 April 2014 09:16
To: [log in to unmask]
Subject: Re: The Value & Impact of Data Sharing & Curation

On the use of subject repositories I look at BBSRC Data Sharing document and it had the statement 

'Researchers should therefore ensure they retain a local copy of any data submitted to third party resources.'

This seems to imply that even if we upload to a subject repository we should also keep an in house copy

Have I misread - is this the same for other funders?

Ricky

Tel: o289o973955
Information Services
71 University Road
Queen's University Belfast
Belfast BT7 1LP


-----Original Message-----
From: Research Data Management discussion list [mailto:[log in to unmask]] On Behalf Of Anna Clements
Sent: 03 April 2014 09:06
To: [log in to unmask]
Subject: Re: The Value & Impact of Data Sharing & Curation

Many thanks for the info, Neil  and certainly 4C should help the sector going forward and I appreciate that it's a complex area and we have to take costs and benefits into account. 

Keeping things simple (I hope!) TCO for in house long-term storage includes 1. capital costs of storage incl replacement cycle
2  cost of backup and DR (here we can take data loss/corruption risk and recovery needs into account)
3 housing & power costs
4 staff costs

Is this how others have split out the costs?

For example, 

Anna

______________________________________________________
Anna Clements | Head of Research Data and Information Services

University of St Andrews Library | North Street | St Andrews | KY16 9TR|
T:01334 462761 | @AnnaKClements

________________________________________
From: Research Data Management discussion list [[log in to unmask]] on behalf of Neil Grindley [[log in to unmask]]
Sent: 02 April 2014 23:34
To: [log in to unmask]
Subject: Re: The Value & Impact of Data Sharing & Curation

Hi Anna,

"The biggest issue I face right now is trying to estimate costs - even for such things that should be straightforward such as the data storage infrastructure. I know there has been some work on this and also the 4C project is underway" ...

Yes, that's right - the 4C Project is looking at costs and related issues and is aiming to put a bunch of resources together that will help people to get a better grip on the economics of digital curation. I stretch the concept to 'economics' because we are taking the view that you shouldn't just focus on the cost of long term info management. You have to factor in benefits; risk (appetite for); value (why is the data important); and sustainability issues (how efficiently setup are you to do digital curation?) to really get a true understanding of cost. We are referring to these and other cost-related concepts as 'indirect economic determinants'. We've written about this on the 4C website ...
http://bit.ly/PP2QYb

But enough already of the 4C approach and back to your question.

"How do I estimate costs - even for such things that should be straightforward such as data storage infrastructure?"

It's a good question. I note that John Milner has put in a couple of comments to this thread already and I feel emboldened to encourage him to dig out the notes he circulated from a Jisc-funded storage workshop that Simon Hodson organised in February last year. There seemed to be a bit of consensus emerging then about some average costs of storage in the university environment. 450 per Tb per year for active data. Approx 1000 per Tb per year for multiple copy resilient storage. For archive data an 'in perpetuity' cost of around 5000 per terabyte ...

But some suppliers and some institutions can do better than this.

It's been interesting as part of the 4C work to map this onto other sources.

Hitachi Systems - http://blogs.hds.com/david/2012/03/big-data-storage-economics-case-study-1.html
Different numbers based on different scenarios ...

An organisation who kindly stepped forward to help the 4C Project (whose business is long-term preservation):
Different tiered storage options ranging from:
873 per Tb per year (2011)
To
11,303 per Tb per year (2011)
Depending on how quickly users want to access the data

And then a meeting at CERN with physicists who look after vast amounts of data who looked a bit puzzled when I quoted them some assumptions about the long term cost of data storage. "Why does it cost you that much?" they asked. (Tip ... If you want to know about efficient data storage and really good ways to migrate software - ask CERN).

If we want to know what data storage will cost us in 5 or 10 years time, we should ask the people who are already doing it. CERN, the Internet Archive, the Church of Latter Day Saints.

We have done or will do this but what we also want to do - and what we think is going to help most - is to set up a Curation Costs Exchange.

http://4cproject.eu/community-resources/outputs-and-deliverables/d2-8-3-3-curation-costs-exchange-ccex-in-progress

I'm going to shamelessly take this opportunity to ask all of you on this list to help with this. Please look out for an announcement that the CCEx is up and functioning and if you can, get some costs data into it.

All the best

Neil


Neil Grindley
Head of Resource Discovery
T 0203 006 6059
M 07816 277 573
Skype neil.grindley
Twitter neilgrindley
Brettenham House, 5 Lancaster Place, London, WC2E 7EN jisc.ac.uk Jisc is a registered charity (number 1149740) and a company limited by guarantee which is registered in England under Company No. 5747339, VAT No. GB 882 5529 90. Jisc's registered office is: One Castlepark, Tower Hill, Bristol, BS2 0JA. T 0203 697 5800. jisc.ac.uk



-----Original Message-----
From: Research Data Management discussion list [mailto:[log in to unmask]] On Behalf Of Anna Clements
Sent: 02 April 2014 19:30
To: [log in to unmask]
Subject: Re: The Value & Impact of Data Sharing & Curation

Thanks, both - and I probably agree overall.

The biggest issue I face right now is trying to estimate costs - even for such things that should be straightforward such as the data storage infrastructure. I know there has been some work on this and also the 4C project is underway ... But does anyone have a costing model they would be willing to share particularly comparing local I house storage to 3rd party provision. (I am aware of the costing model for Arkivum) Particularly given that we are also dealing with considerable uncertainty on 1. How much data we have and growth rate 2. How much we expect to be able to store in external data centres and 3. How much we think will be open access and therefore easily accessible and shareable cf what we expect to be restricted / by request access - so scalability is a key factor.

Anna


Anna Clements
Head of Research Data and Information Services University of St Andrews St Andrews, Fife,KY16 9AL @annakclements

> On 2 Apr 2014, at 18:56, "Kevin Ashley" <[log in to unmask]> wrote:
>
>> On 02/04/14 18:21, Anna Clements wrote:
>> Thanks, Rachel and for the summary - will read in detail but initial 
>> question is ...
>> Is this report then evidence for the need for (inter)national & 
>> subject data centres to be developed where none exist .. and so maybe 
>> remove the need for institutions to do this themselves - duplicating costs and effort?
>
> My opinion:
>
> The answer is yes to the first part, and no to the second. Yes, it is 
> evidence that the existing data centres deliver value and hence can 
> support the case for more of them to exist. But even if funders take 
> that up with enthusiasm, it will take some years for it to happen and 
> there will still be areas of research that aren't covered. There will 
> thus be a role for institutions for the foreseeable future. These 
> questions were examined in the UKRDS report back in 2010 and I don't think the picture has changed substantially since then.
>
> As John Milner said, the solutions may involve regional consortia 
> and/or some mix of commercial suppliers for some of the 
> infrastructure, but intellectual custody will still remain the responsibility of institutions themselves.
>
>

Janet(UK) is a trading name of Jisc Collections and Janet Limited, a not-for-profit company which is registered in England under No. 2881024 and whose Registered Office is at Lumen House, Library Avenue, Harwell Oxford, Didcot, Oxfordshire. OX11 0SG. VAT No. 614944238