BCS Kingston & Croydon - Oxfam Case Study - Feb 2013


Published on

Slides from talk by Stewart Marshall and Graham Oakes to BCS Kingston & Croydon Branch on 12 Feb, 2013.

Published in: Technology
  • Be the first to comment

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide
  • Stewart
  • StewartWorks with others to overcome poverty and suffering: Humanitarian, development, campaigns5000 staff. 60 countries. 22000 volunteers. 1M supporters. 700 high-street shopsWeb: Supporter engagement, campaigns, actions - Online shop: new, donated, unwrapped (first UK online charity shop, 2007)Donations: regular, single - Professional publications
  • Stewart:The Problem (1) – Tangled systems due to organic growth of web presence over last 8-10 years. (30 servers in hosting environment; with tangled web of applications across them; no DR; ill-defined SLAs; poor availability, largely due to complexity & lack of SLAs)
  • And they had a larger operational problem: Large, unpredictable peaks in traffic. Can handle be reducing website to single page, but this limits ability to build relationship and follow-up with supporters. Talk through the 3 demand regimes. Private car – hire car – taxi model.
  • Graham
  • Graham – The Procurement:Use Gartner to shortlist to 6 vendorsSend spec to trigger collaborative procurement (6 weeks response, with 2-3 conference calls + full day workshop per vendor)Shortlist to 3 for presentationsShortlist to 2 for “BAFO”
  • Stewart – Cloud is about economics, not technologyThe technologies underpinning the cloud are now relatively tried and tested. Virtualisation has been around for years (decades if you remember mainframes). Aspects of the technology, e.g. APIs for capacity management and provisioning, are still stabilising but these are second order elements. The core is there and working.The biggest challenge we faced was understanding the economics of price and capacity elasticity – how does the overall cost of service change under different demand patterns and pricing models? This is also the biggest opportunity. Get the model right, and Oxfam can reduce costs for normal operations while still being able to ramp up to take the opportunities that demand spikes bring. To understand these economics we explored several demand and pricing scenarios. It’s here we learnt our second lesson.
  • Graham – Standards are still emerging.People talk about utility computing and compare cloud to the electricity grid. There’s truth in that, but only if you look at the electricity grid of the 19th century. Some Infrastructure-as-a-Service vendors are running AC; some are running DC. They’re all running different voltages. And it costs less to run a kettle than two-dozen lightbulbs, even though you’re drawing the same amount of power either way. This makes it very hard to compare different vendors’ offerings.
  • Stewart – The market for cloud capacity is immature.Oxfam isn’t a technology company. It doesn’t want to own servers any more than it wants to own power generators. A utility model is just what Oxfam needs. We ran the procurement on this basis: we specified the amount of capacity we wanted, saying as little as possible about technology.So, we built a model of our requirements for compute capacity (measured in terms of CPU cores, RAM, storage and bandwidth). We identified the capacity we would need in a typical month; how this might grow for seasonal demand (Oxfam is a major retailer, after all); what spikes we might need to deal with. We mapped this onto the availability levels needed to support the various applications running on the infrastructure.After all this, we found that we needed to spend a lot of time helping the vendors understand these requirements. Most of them needed to turn our capacity model into a technical configuration before they could price it. They were still thinking in terms of servers and blades, not computational capacity. And the only way we could compare different vendors’ pricing was to examine their technical configurations – no-one would directly assure specific levels of capacity and availability without also specifying the technology.We’re convinced we took the right path – we want capacity, not servers, so this is what we specified. But for now, despite the hype, the market is really selling servers.
  • Graham – Manageability matters as much as capacity.The cost of service depends as much on the number of virtual machines you’re running as it does on the overall amount of compute capacity. This is logical when you think about it – each VM needs to be managed – so this needs to be built into the requirements models.This is also going to complicate ongoing management of the infrastructure. It’s not enough to just add capacity as we need it: the way this capacity maps onto VMs will also drive costs. This affects each application differently due to their different licensing and technical models.
  • Stewart – Application licensing constrains deployment.Some of our web applications are designed to scale horizontally: the best way to deploy them is as a swarm of small VMs. However, they are licensed on a per-server basis, making this very expensive. For now we’re deploying a small number of larger VMs, and hoping our software vendors will eventually catch up with the cloud.
  • Graham – Don’t get hung up on a name.Does it matter if it’s “public cloud” or “private cloud”? All the solutions we looked at were based on a hybrid cloud, but some of them emphasized dedicated kit on a private cloud while others emphasized shared kit in a public cloud. We agonised over whether this mattered – whether one or other gave greater flexibility, for example. In the end, we came back to economics: once they met the basic requirements for capacity, availability, security, etc, then which gave the best overall cost of service? We didn’t care what label it carried.
  • Stewart – Negotiation is a key part of the projectIt takes a lot of time to get the price & terms rightIt helps to have negotiation specialists (the vendor does!)Sticking points – data liability, availability, ethical standards
  • Graham – you need internal skillsEven if vendor is providing end-to-end service, you need skills to manage them, and to handle interfaces with other vendors (apps, etc)
  • Stewart – Cloud opens up other internal issuesRelationship between IT and business units it provides a service for – now based on a utility service model, not bespoke servers; scope to rethink the relationship – how it works & what it coversWay infrastructure/ops teams supports projects – teams can set up their own VMs, but need to control sprawlArchiving – storage costs are transparentChargeback models – transparency of costs; can bring platforms up & down easilyLifecycle models – change when can bring platforms up/down easily
  • Stewart:Were aiming to control the lifecycle, but it’s still very project based – no concept of an ongoing application/product lifecycle (and no tools beyond spreadsheets to manage any of this)
  • Stewart
  • Stewart
  • Graham
  • Graham
  • Graham
  • Graham
  • Stewart
  • Stewart
  • Summary
  • Who I amIndependent consultantDo 2 things – help set up project (untangle complexity); help keep in touch with what’s going onUnusual perspective on assurancePortfolio of mid-size projects rather than single large programmeDifferent twists, but aligns to where many organisations are at, so will share experienceAgenda
  • BCS Kingston & Croydon - Oxfam Case Study - Feb 2013

    1. 1. Moving to the Cloud Oxfam’s JourneyMoving to the CloudFeb 2013 1
    2. 2. Contents Oxfam The Problem The Project Lessons Learned Discussion – what does it all mean?Moving to the CloudFeb 2013 2 kcdsTM
    3. 3. Moving to the CloudFeb 2013 3
    4. 4. Moving to the CloudFeb 2013 4 dogbomb
    5. 5. Burst PeakDemand  Base Time  Moving to the Cloud Feb 2013 5
    6. 6. Solution – variable capacity Best efforts Business hours only 7x24, highly responsiveMoving to the CloudFeb 2013 6 kcdsTM
    7. 7. TheProjectMoving to the CloudFeb 2013 7 Lachlan
    8. 8. Lessons LearnedMoving to the CloudFeb 2013 8 kcdsTM
    9. 9. Cloud is about economics, not technologyMoving to the CloudFeb 2013 9 dogbomb
    10. 10. Standards are still emergingMoving to the CloudFeb 2013 10 courtneyBolton
    11. 11. The marketfor cloud capacity is immatureMoving to the CloudFeb 2013 11 Sudhamshu
    12. 12. Manageability Matters as much as CapacityMoving to the CloudFeb 2013 12 clagnut
    13. 13. Application licensing constrains deploymentMoving to the CloudFeb 2013 13 RickC
    14. 14. Don’t get hung up on a nameMoving to the CloudFeb 2013 14 Олександр
    15. 15. Negotiation is akey part of the projectMoving to the CloudFeb 2013 15 net_efekt
    16. 16. You need internal skillsMoving to the CloudFeb 2013 16 Sister72
    17. 17. Cloud opens up other internal issuesMoving to the CloudFeb 2013 17 jronaldlee
    18. 18. a) We now had clearer visibility of the Discovery application portfolio Definition Ideation Support Launch Build b) We could now set up new servers / environments very quickly (creates sprawl if unmanaged, or an opportunity if managed) GovernanceMoving to the CloudFeb 2013 18 dogbomb
    19. 19. Application Portfolio high Differentiating apps, deliver Experimental How differentiating / competitive apps, building advantage now for the future innovative? (e.g. community (e.g. Enabler) fundraising) Commodity Commodity apps, critical to apps, support business operations operational (e.g. efficiency email, finance, HR) (e.g. intranet) low core peripheral How critical to operations?Moving to the CloudFeb 2013 22
    20. 20. Application Meta-Lifecycle high Differentiating Deliver deliver apps, Experiment Experimental How differentiating / competitive invest apps, building advantage now for the future innovative? (e.g. community (e.g. Enabler) fundraising) stabilise Commodity Commodity apps, critical to apps, support business operations operational (e.g. efficiency Commoditise email, finance, HR) Minimise (e.g. intranet) low core peripheral How critical to operations?Moving to the CloudFeb 2013 23
    21. 21. Development Style Experiment Deliver Commoditise MinimiseCycle Time: Days Weeks Months MonthsRelease Immediate 4-6 per annum 1-2 per annum 1 per annumCadence:Team: Small, co- Scrum Scrum? Outsourced located, x- function SWATApproach: Kanban Scrum Scrum? Spec & buyTransition to Extract key Re-architect: Retire Retirenext style: integrate more features, harden, and tightly with core integrate with apps & reduce operational cost core apps Moving to the Cloud Feb 2013 24
    22. 22. DiscussionMoving to the CloudFeb 2013 26 kcdsTM
    23. 23. Summary Cloud is a viable technology option The market is not yet mature You need skills to manage cloud vendors Cloud changes internal relationships and processesMoving to the CloudFeb 2013 27
    24. 24. Thank Yougraham@grahamoakes.co.ukSMarshall@oxfam.org.ukMoving to the CloudFeb 2013 28
    25. 25. Graham Oakes Ltd Making sense of technology…  Many organisations are caught up in the complexity of technology and systems.  This complexity may be inherent to the technology itself. It may be created by the pace of technology change. Or it may arise from the surrounding process, people and governance structures.  We help untangle this complexity and define business strategies that both can be implemented and will be adopted by people throughout the organisation and its partner network. We then help assure delivery of implementation projects. Clients…  Cisco Worldwide Education – Architecture and research for e-learning and educational systems  Council of Europe – Systems for monitoring compliance with international treaties; e-learning systems  Dover Harbour Board – Systems and architecture review  MessageLabs – Architecture and assurance for partner management portal  National Savings & Investments – Helped NS&I and BPO partner develop joint IS strategy  The Open University – Enterprise architecture, CRM and product development strategies  Oxfam – Content management, CRM, e-Commerce  Thames Valley Police – Internet Consultancy  Sony Computer Entertainment – Global process definition  Amnesty International, Endemol, tsoosayLabs, Vodafone, …Moving to the CloudFeb 2013 29