Cloud Computing Workshop - Presentation Transcript

1. Cloud Computing Primer: Steps for using the Cloud in Your Museum Ari Davidow – jwa.org Charles Moad – imamuseum.org Robert Stein - imamuseum.org
2. wikipedia on cloud computing via wordle.net
3. CLOUD COMPUTING CLOUD APPLICATIONS UTILITY COMPUTING
4. Cloud Applications ... eliminate the need to install and run the application on the customer's own computer, thus alleviating the burden of software maintenance, ongoing operation, and support.
5. Cloud Applications
6. Utility Computing … a style of computing where scalable and elastic IT-related capabilities are provided as a service to external customers using Internet technologies. 
7. Utility Computing
8. Buzz Worthy Search Trends for “cloud computing” Via Google Search Trends
9. Source “Cloud Computing Gains in Currency”, Pew Research, May 2008 http://pewresearch.org/pubs/948/cloud-computing-gains-in-currency 69% of Americans use cloud computing services
10. Gartner’s Hype Cycle for 2009 image courtesy of gartner.com
11. 21% of companies are piloting SaaS applications up from 18% last year – Forrester, Feb 2009
12. In Forrester’s List of the Top 15 Technology Trends
13. State of Cloud Computing
    • Forrester feels that cloud computing is one of the  Top 15 Technology Trends  and that it warrants investment now so you can gain the experience necessary to take advantage of it in its many forms to transform your organization into a more efficient and responsive service provider to the business
    • - Forrester, October 13, 2009,
    “ http://blogs.forrester.com/it_infrastructure/2009/10/cloud-computing-belongs-on-your-3year-roadmap.html
14. Gartner’s #1 Strategic Technology Area for 2010
15. State of Cloud Computing Cloud computing isn't going to be vapor much longer… It's complicated, poses security risks, and computing technology companies are latching onto the buzzword in droves, but the phenomenon should be taken seriously… -Gartner - October 20,2009 http://news.cnet.com/8301-30685_3-10378782-264.html “
16. Concerns about SaaS
17. Pros of Cloud Computing
    • Fast Deployment
    • Lower cost / No Capital Expense
    • Reduced IT maintenance
    • Elastic and Unlimited Scalability
    • Energy Efficiency
    • Reliability (Service & Data)
    • Better Resource Utilization
18. Cons of Cloud Computing
    • Information Security
    • Physical Security
    • Long Term Offline Storage
    • Bandwidth Bottleneck
    • Potential Vendor Lock-in
    • Lack of control during downtime
19. Amazon Web Services (AWS) Overview Amazon Web Services (AWS) Infrastructure Services Elastic Compute Cloud (EC2) Simple Storage Service (S3) SimpleDB Simple Queue Service (SQS) Elastic Block Store (EBS) Elastic MapReduce CloudFront Content Delivery Network Relational Database Service (RDS) Virtual Private Cloud
20. How to make choices about Cloud Computing
    • What sort of security requirements fit your data?
    • How granular is the information you’re working with? (documents, images, video?)
    • Where are your likely performance bottlenecks? (compute, bandwith, latency)
    • What is your IT staff like? (small but flexible, large)
21. Jungle Disk
22. Jungle Disk - $20 / www.jungledisk.com
23. Requires
    • Amazon S3 account, and the requisite keys:
    • JungleDisk software installed
24. Usage
    • Backs up at scheduled times
    • Can back up more than one machine, or to more than one backup set
    • The first backup may take days – or longer. No problem. The software gracefully goes to sleep when you shut down or hibernate; resumes upon waking until done
    • Can retrieve files using drag/drop interface using pull-down to set the date of the view from which you wish to retrieve (i.e., let me see the files as they were on July 7, 2008).
    • Retrieves files gracefully and quickly
25. Converse Example
26. IMA’s SAN
    • IMA Purchased 32TB of EMC SAN in 2006
    • 16TB local and 16TB at an offsite co-location facility
    • Due to growth in Collection Photography, Video, and Conservation Imagery that space is all but full!
27. IMA’s SAN
    • Benchmark Growth Rate
    • Total Current Size – Initial Size / 36 months
    • Ballpark Rate of 142 GB/month
    • Yielding 13.9TB estimated in the next 4 years
28. IMA’s SAN
    • 16TB Onsite + 14TB AWS = $164,544
    • 16TB Onsite + 16TB Colo = $94,200
29. Hidden Costs
    • DS3+Colo = $96,000 (3 yr commitment)
    • Maintenance and administration of servers (2 FTE’s)
    • As replicated backup for super large file-systems time for restore would be huge!
30. AWS Tools
31. Elasticfox & S3Fox
32. EC2 Console
33. Moving Drupal to the Cloud: Step by step
34. Introducing the EC2 Console
35. Creating a Key Pair
36. Creating a Security Group
37. Selecting a Starter AMI
38. Launching an AMI
39. Connecting to Your EC2 Instance
40. Creating an EBS Volume
41. Configuring Apache and MySQL
42. Setting up Drupal
43. Bundling an AMI
44. ami-764bab1f
45. Fedora for DAM
46. Fedora as a testbed on AWS Project Goals: External vendor to create least-possible Fedora instance to enable preservation work
47. Summary
    • Create on AWS and hand over instance when done
    • When dev site is completed, create “Amazon Machine Instance” and check into Subversion
    • Document installation and everything else in wiki
    • We create new instance from checked-out AMI
    • This ensures that we have maintainable code that we can get up and running, before developer moves on
48. What we did
    • Original server created using developer’s favorite Linux
    • We use CentOS, so when we checked out the AMI, we recreated running under CentOS, bundled new AMI to S3
    • AMIs can be independent of the underlying OS
49. Bugs
    • Our repository, which consists of lots of very large files, uses a unix filesystem called XFS
    • XFS supports very large volumes better the usual filesystem, and supports real-time snapshotting of huge file-systems
    • AWS updated CentOS and broke XFS
    • We (actually, our webmaster) rebuilt kernel to work around AWS CentOS bug
50. Other Gotchas
    • An EC2 instance doesn’t preserve state
    • When you restart, it restarts from scratch
    • All config changes, anything else that was done and saved to the previous instance is gone
    • So, you use EBS, which acts something like a network drive (think NetApps)
    • You purchase blocks of EBS space at a time, but it is cheaper than S3 per GB/Month
    • This is different from S3 storage where you pay only for what you consume
51. AMIs
    • Amazon Machine Instance
    • Sort of like a “ghost”ed server image
    • Amazon (and others) provide lots of AMIs to work with
    • AMIs can be public or private
    • You can use different AMIs on different servers in your AWS setup
    • “ bundling” is the AWS term for saving that AMI with your modifications for future use.
    • We store AMIs on S3; could also use EBS
52.  
53. Lessons Learned
    • We liked AWS so much, and saved so much money, that we have now moved all of our web services to AWS.
    • Our website used to cost us $1200/mo. It has added about $450/mo to what we already pay for the Fedora instance – about $900/mo total.
54.  
55.  
56. Rightscale – basic services free www.rightscale.com
57.  
58.  
59. AWS Infrastructure Amazon Web Services (AWS) Infrastructure Services Elastic Compute Cloud (EC2) Simple Storage Service (S3) SimpleDB Simple Queue Service (SQS) Elastic Block Store (EBS) Elastic MapReduce CloudFront Content Delivery Network Relational Database Service (RDS) Virtual Private Cloud
60.  
61. Scalabble
62. Video Processing
63. Total Monthly AWS
64. Monthly Bill for ArtBabble.org Web Server
65. Monthly Bill for Wowza Video Server(s)
66. AWS Bill - CloudFront
67. AWS Bill – EC2
68. AWS Bill – S3
69. AWS Bill - Wowza
70. The Numbers (so far)
    • 150,000 video views
    • (168k visits / 576k pages)
    • 81,000 note clicks
    • 1:3 of the notes expanded
    • 22,400 views of “Behind the Babble”
    • 25,015 views of most popular YouTube video posted Feb. 1 st , 2008
    • 5,000 registered users
    • 44% signed up using OpenID (but didn’t realize it)
71. The Numbers (so far) for geeks
    • 112 hours of video processed
    • 525 videos
    • 1700 instance hours
    • At a cost of ~ $0.65 per video
    • April 1 st – October 30 th
    • 167,000 visitors
    • From 166 countries
    • April 1 st – September 31 st
    • 1.1TB of web data transferred out
    • At a transfer cost of $200
    • 1.25TB of video streamed
    • At a transfer cost of $250
    • 11 Mbps average transfer on embedded videos
    • IMA just upgraded to 5Mb pipe Fall ‘08
72. QUESTIONS?