Archive for the ‘cloud computing’ Category

Saturday, June 26, 2010

8:00 a.m. – 12:00 p.m.

Washington Convention Center – 143A

 

Twitter hashtag – #litacloud

 

This session was presented as a panel discussion followed by a lightning round followed by another panel discussion. Some of the panelists included:

 

Marshall Breeding, Vanderbilt

Karen Coombs, OCLC

Terry Reese, Oregon State University


 

Cloud computing: characteristics

 

NIST Definition – http://csrc.nist.gov/groups/SNS/cloud-computing/

on-demand self-service

broad network access

resource pooling

rapid elasticity

measured services

 

Gartner definition

service based

uses internet technologies

shared

scalable and elastic

metered by use

 

Cloud computing is not merely a delivery method.

 

Cloud based software-as-a-service

software delivered via the cloud

 

Cloud based platform-as-a-service

platforms (LAMP stack, Ruby on Rails stack, etc.) delivered via the web

 

Cloud based infrastructure-as-a-service

 

www.heroku.com – Heroku provides an online development/testing platform

Requires a Ruby on Rails application that you’re uploading.

Heroku is using Amazon’s online services as the infrastructure for their own platform as service.

 

Marshall Breeding, Vanderbilt University

 

Continuum of Abstraction

Locally owned and installed servers

Co-located servers

Co-located virtual servers

Web hosting

Server hosting services

Application service provider

Software as-a-service

Platform-as-a-service

 

Cloud computing – formal definitions

Highly abstracted computing model

Utility model

Provisioned on demand

Scaled according to variable needs

Discrete virtual machines

Compute cycles on demand

Storage on demand

Elastic – consumption of resources can grow and contract on demand

 

Hosting Services

Web Hosting

Web site only

Standard support for PHP, Perl, and other dynamic page generation

Dedicated Server

Appropriate for applications that have not been tested and deployed in a virtual environment

 

Advantages

Increasing opportunities to eliminate local servers and tech support

All of Serials Solutions’ offerings are delivered as software-as-a-service

 

Liblime

Liblime Enterprise Koha deployed in Amazon EC2

LAMP stack implemented on Virtual Machine Image

Ability to meet larger site requirements through high-performance cloud-delivered platform

 

Karen Coombs, OCLC

 

Cloud, Community, Collaboration

 

Collaboration in the Cloud

Infrastructure and tools exist to facilitate better collaboration across libraries

Beak down boundaries between developers in different libraries

Infrastructure alone is not enough. We have to change the ways libraries collaborate.

 

Transparency and the Cloud

Documentation of cloud application’s infrastructure and capabilities

Web services to as many aspects of the application as possible

Standards based systems (web standards, not library standards)

Blackboxes in the cloud diminish the real power of the cloud – collaborative innovation

 

Software in the Cloud

Ability to develop in potentially a device and platform independent way

computers, smart phones, single-purpose devices like e-readers

Creates opportunity for geater scalability

Relives the burden of installation and updates

Shared software, libraries, and infrastructure

Don’t have to develop all of these core services locally

 

Software Development and the Cloud

Cooperative development

Open source projects have been doing this for some time

Shared development effort

Ability for institutions and individuals to participate in different ways

Crowd Sourcing

Testing

Coding

Systems are designed in a modular fashion to allow developers to extend them.

 

Terry Reese, Oregon State University

 

Moving Library IT to “International Waters”

 

International waters – the idea that in some environments, a completely different set of rules apply.

 

Shared IT Resources Are Hard

IT resources (staff and hardware) represent a finite and expensive resource

Disks are cheap until you get a lot of them

Server cycles are expensive because they are finite within a given infrastructure

Possibly the biggest barrier is organizational

While projects will have multiple partners, one partner has the responsibility for managing and support the infrastructure.

 

With cloud computing you can move the project outside the organizational bureaucracy and into international waters where projects can function unencumbered.

 

With freedom comes options:

Add new partners at will

Partners determine how resources are managed; if you change you mind, that’s fine.

Allows a project to “think bigger” because most cloud resources will scale almost at will.

 

DuraCloud – DuraSpace

A hosted service and open technology to help organizations and end users effectively utilize public cloud services.

Built upon existing cloud services.

The service can work on Amazon, Atmos, Sun, Rackspace, and other cloud services.

LOCKSS in the cloud based on DuraCloud.

 

Chronopolis Project – designed primarily as a preservation storage system

Chronopolis Tools also monitors files and does auditing.

IRODS

 

TerraPod – digital video library

Allows you to outsource upload and data creation to the creators of the content.

 

Disadvantages

Data in the cloud – loss of control

Terms of service

API lag

Varying support

Advertisements

As a general rule, I don’t like my data in the cloud. Let me go ahead and get that out there so my prejudices are fully disclosed. I use it because I must, but I don’t like it. The primary reason I don’t like it is a practical one: data in the cloud means that you have to be able to connect to the cloud, and far too often, I can’t. Living and traveling in a largely rural area, network coverage is far from ubiquitous. And no network coverage sometimes translates into no data. Setting aside security concerns and the additional hook into my devices that each application seems to want, connectivity is my main concern.

Until recently, that is. Several months ago, a friend recommended Evernote. There are a lot of things I like about Evernote. I like the clipping feature, and I like the IDEA of being able to access my notes from many locations. Unfortunately, I recently lost a lot of data to Evernote. As in POOF – gone forever.

I was recently on vacation, and I started making some travel notes with Evernote. As luck would have it, my vacation area didn’t have great coverage, and Evernote had a lot of trouble every time it tried to sync with the Evernote server.  Initially the result was relatively benign but nevertheless annoying: sections of content would be duplicated several times throughout the document. I had to scroll up and down, find the old stuff, find the new stuff, delete the duplicate stuff, and save it again. Annoying, but doable.

Until the very last day. Over the course of my vacation I spent several hours making notes so that I could remember specific details about the trip. And then the last day everything disappeared. Well . . . almost everything. I still have a blank document with a title, but that’s all. Everything else – all of my CONTENT – is gone. It happened after one of those periods when my local application was trying to phone home. Apparently something went awry, it couldn’t connect properly to the Evernote server, and Evernote inexplicably thought that I wanted to delete all of my content. Grrrrr. Double-grrrrr even.

These were just vacation notes, true. It wasn’t critical work/medical/financial information. But the principle is the same. I relied on the cloud. The cloud ate my data. The cloud failed. The scary thing is that more and more companies are pushing data for mobile devices into the cloud. While the idea is a good one, the execution is everything. If you’re going to eat my data, I really can’t trust your service. Moreover, I can’t in good conscience recommend it to others.