Practical Research 1: Lesson 8 Writing the Thesis Statement.pptx
UK LOCKSS Alliance: Today’s scholarly content, secured for tomorrow
1. UK LOCKSS Alliance
Today’s scholarly content, secured for tomorrow
Adam Rusbridge
UK LOCKSS Alliance Coordinator
EDINA, University of Edinburgh
8th March 2012
Trust and eJournals Workshop, London
2. Summary
• LOCKSS: Digital equivalent of the physical shelf
• Sufficient rights to access content as needed
• Financial control and governance over systems
• Automate preservation functions where possible
• LOCKSS provides generic preservation capacity
• Customise the distributed architecture according to
community needs
• Modeling the total cost of long-term storage
www.flickr.com/photos/guitarlogy/5387073471/
3. Community Action for Assured Access
A co-operative organization to ensure continuing sustainable access to scholarly work over the long term.
UK libraries are collaborating to build national ‘network level’ infrastructure and to coordinate the preservation of electronic
material of local and UK interest. (since 2008)
17 member institutions Steering Committee directs activity (next
De Montfort University meeting May 2012)
King’s College London
London School of Economics
Natural History Museum Phil Adams (De Montfort University)
Open University Lisa Cardy (London School of Economics)
Royal Holloway, University of London
University of Birmingham
Geoff Gilbert (University of Birmingham)
University of Edinburgh Tony Kidd (University of Glasgow)
University of Glasgow
University of Hertfordshire
Liz Stevenson (University of Edinburgh)
University of Huddersfield Lorraine Estelle (JISC Collections)
University of Newcastle Upon Tyne
University of Oxford
Peter Burnhill (EDINA)
University of Salford Adam Rusbridge (EDINA)
University of St. Andrews
University of Warwick
University of York
Support Service at EDINA provides underlying coordination, support and development
JISC Collections organises membership subscriptions and gives guidance and support
JISC prompted the initial project led by the Digital Curation Centre (2006-08)
4. Technical Infrastructure
- Preserves content as published
- Preserve the record: web archiving
- Fetches content from a server
- Preserves integrity
- Audit protocol to prevent damage
- Tamper resistant
- Avoids single point of failure
- Distributed network to avoid points
of failure
- Model on success of print collections
(and operation of the library)
5. Technical Infrastructure
- Preserves content as published
- Preserve the record: web archiving
- Fetches content from a server
- Preserves integrity
- Audit protocol to prevent damage
- Tamper resistant
- Avoids single point of failure
- Distributed network to avoid points
of failure
- Model on success of print collections
(and operation of the library)
6. Technical Infrastructure
- Preserves content as published
- Preserve the record: web archiving
- Fetches content from a server
- Preserves integrity
- Audit protocol to prevent damage
- Tamper resistant
- Avoids single point of failure
- Distributed network to avoid points
of failure
- Model on success of print collections
(and operation of the library)
7. MetaArchive
• A distributed digital preservation solution depends on a
collaborating set of institutions agreeing to preserve each other’s
content.
• Requires central coordination; shared enthusiasm, resources and benefit
• Successful models initiated where community / shared need already in place.
• MetaArchive is a cooperative not a vendor (conceived 2004)
• Goal is not to make profits, but to improve each member's situation.
• Distribute across geography: diversify funding, politics, economy
• Replicate content, lower barriers of entry
• Educopia Institute - non-profit administrative organisation
• Coordination role; arrange legal agreements and commitments to
preserve member content
• Sustained by affordable cooperative fee memberships set by
members
• Supplemented by grants and contracts
8. Costs
• Equipment
• Each institution required to contribute a server to the network
• As of June 2011: $4,600 for a 16TB machine
• Staffing
• 2% of a systems administrator’s time
• Administrator/point of contact
• Software engineer who preps content for ingest
• (latter two roles needed for outsourced solution).
• Storage
• $1.00/GB/year for content stored in the network.
• ‘Conspectus’ to organise where content is stored
9. Tiered Membership
• Sustaining Members: $5,500/year
• Leadership, development, governance
• Preservation Members: $3,000/year
• Benefit from shared preservation model
• Collaborative Members: (varies, but e.g. $4,000/year for 20)
• Consortia that share a server, and so look like one organisation.
• Allows existing consortia to preserve co-hosted content for a
fraction of what it would cost to do so as individual members.
10. PLNs in the UK: Member Survey
• Share resources & responsibility, build community, keep costs
low
• Preservation policy
• Content and Collections
• Organisational architecture
• Costs and Resources
11. Initial Conclusions
• Survey response rate: 50% of members
• Institutions seeking affordable solutions to digital preservation.
• e-preservation strategies have yet to be developed.
• Extent of digital assets requiring preservation unknown
• Systematic audits have not yet been carried out.
• Prefer architecture where content is stored at more than one location
• However a fully distributed approach was not favoured.
• Mixed enthusiasm for a PLN
• Need to demonstrate PLN is low-cost and sustainable
• Need clear and demonstrable financial benefits
• Need a shared interest in preserving a particular body or type of content.
• Difficult to gain acceptance and commitment without these benefits
• Moving forward: establish a UK PLN, or join the MetaArchive as a
Collaborative Member?
12. Pricing of Long-term Cloud Storage
• David Rosenthal has been looking at cost models for long-term
Year Cost
storage: http://blog.dshr.org per GB
• Does it make economic sense to store data in the cloud, in the 2002 $3.98
long-term. 2004 $1.25
• Kryder's Law, 30yr history of exponential increase in disk 2006 $0.64
capacity at roughly constant cost.
2008 $0.29
• The cost of storing bits for the long term depends on current
price and how fast it is dropping. 2010 $0.08?
2012 $0.06?
• How long can we expect Kryder's Law to continue?
• Indications that Kryder's Law is slowing down
• 4TB disks now available, but slower than expected.
• Driver for 3.5" disks has been desktop PCs. Volume
market is now 2.5" disks: same curve but higher price/
byte.
• By 2020, ought to have 14TB 2.5" drive @ $40
• Consumers may prefer a 2TB 1" drive for $15 and less power
draw
13. Cloud Storage Price History
Provider Price (Year of Current Price % decrease
Launch)
Amazon S3 $0.15/GB/mo $0.125/GB/mo 3%/yr
(2006)
RackSpace $0.15/GB/mo $0.15/GB/mo 0%/yr
(2008)
Windows Azure $0.15/GB/mo $0.14/GB/mo 3%/yr
(2009)
http://blog.dshr.org/2012/02/cloud-storage-pricing-history.html
• Price of cloud storage is dropping around an order of
magnitude more slowly than raw disk prices.
• There is a recurrent cost for storage in the cloud. As
collections grow, will the cost of cloud storage grow more
than if performed locally?
• Research to model total costs over time – local hardware,
maintenance, location, power, bandwidth, staffing.
14. Cost to preserve 8TB
• Starting with S3's current pricing and assuming that it continues to
drop at 3%/yr, the total cost over 4 years would be $41,065.
• DIY: 3 geographically separate complete copies each protected
against double disk failures.
• Three Drobo FS network file servers ($600 each at Amazon)
populated with 5 3TB Hitachi 5400RPM drives ($210 each at
Newegg). Add one spare for each Drobo to cover while failed drives
are returned under warranty. Capital cost of $5580.
• Each Drobo consumes ~70W with all drives active. So we'd
consume 1840 KWh over 4 years. Palo Alto Green rates, cost of $250.
• Stanford experience with Drobos is that almost no attention needed,
but assume staff costs at $50/hr for 1hr/mo/box = $7200.
• The total cost over 4 years would be $13,030: around a third
of the total cost of S3.
http://blog.dshr.org/2012/02/cloud-storage-pricing-history.html
15. Principles of LOCKSS: Building Trusted Archives
• LOCKSS software can be used to provide general, shared
preservation capacity
• Responsibility spread across the community
• Shepherded by strong universities with strong collection
policies
• Further assessment of UK Private LOCKSS Networks
• Model selected depends on scale of content & community
enthusiasm
• Further assessment to understand the total cost of
storage
www.flickr.com/photos/guitarlogy/5387073471/
16. Find out more…
http://www.lockssalliance.ac.uk
a.rusbridge@ed.ac.uk
@EDINA_eJournals