A case study of Nestoria. Discuss the difference between a simple mash-up and a business that runs on mash-ups. Also touch on some challenges we have faced.
6. Milestones
• Start dev April 2006
• Launch London June 2006
• Launch UK Sept 2006
• Launch Spain May 2007
7.
8.
9. Numbers
• In the UK, have 850,000 listings
refreshed continually
• Boils down to about 500,000 properties
• 100,000+ users per month
• 4 programmers
• No dedicated operations staff
12. The Plan
• Talk about some of the technical
decisions we made at Nestoria
• Go over a few challenges that we have
faced
• Hopefully there are lessons to learn
from our experiences
36. Why write tests?
• Find bugs, of course
• Increase confidence to make changes
to unfamiliar systems or libraries
• Promote good design / expose bad
design
37. Design
• Informal process (no UML, minimal
design patterns)
• Concentrate on the things that will really
hurt - e.g. i18n
• Not everything has to be an object
39. Outsource
• If you can spec it faster than you can
write it, consider farming it out
• Many, many site full of hungry coders
• Have had good results with
scriptlance.com but there are lots of
others out there
43. Virtualization?
• Sounds like voodoo to me
• Extra layer to worry about
• Easy enough to do everything in user
space
44. Wigwam
• 12 subdomains
• Live data
• 22 services
• Many playpens per box
• Releases several times a week
45. big ISP vs. small ISP
• Bigger - more consistent
• Smaller - more personal
• Bigger - discrete choices
• Smaller - continuous advice
• We went small
46. Hardware
• Know your downtime limits
• Be reasonable!
• Load balancing - expensive and
complex - we do it with DNS through a
vendor
• Have a failure plan
49. Logging
• Need your own, and it needs to be
flexible
• Use the free tools as a sanity check
50. Reporting
• An important area of expertise, not an
annoying afterthought
• Glossary
• Every mean needs a variance
• A/B Testing (a.k.a Bucket Testing) is
hard to do right
51. Also…
• Build tools for ad hoc queries, they
>will< get used
• Keep an audit trail of changes
• Effective tools for drilling down on
sessions allows non-techs to help
debugging
65. Geocoding
• Transform address into lat/long
coordinates
• 1,000+ calls a minute
• Very high data variance (few cache hits)
• Royal Mail data for UK
• Teleatlas data for Spain
66. • Albert Villas, Gilbert Mews, LEIGHTON BUZZARD,
• Bedfordshire, LU7 1NF
• Lake Lock Drive, Wakefield West Yorkshire
• Church Lane, London
• C ELMINGTON ROAD, LONDON
• Fountain Road (Flat 2), Edgbaston
• London SE18, UK
• Lee, London, UK
• Free Trade Wharf, 340 The Highway,
Wapping/Limehouse London
• Stunning 3 Storey Contemporary Barn Conversion In
The Popular Village Of Utkinton
• A.4-1 GREAT NORTHERN TOWER,WATSON
STREET,CITY CENTRE,
• MANCHESTER GREATER MANCHESTER
• 15, ASTLEY HOUSE, LONDON
70. Future Projects
• Expose more data via API (e.g.
historical aggregate data)
• Implicit personalisation (the system
learns from users)
• More international expansion
• Systemic testing to improve user
experience