elemental links

Liveblogging: Cloud Computing Expo #9 Mike Hill (IBM)

Mike Hill, VP, IBM Cloud Services on How the CIO becomes a hero again:

‘Cloud Computing will have major impacts to the way in which IT is delivered from the data center to its many clients. For the CIO, this can be perceived as a threat to the way in which they manage data center capabilities. In reality, cloud computing offers a huge opportunity to the CIO who now has the capability to offer more responsive, scalable, and available IT services. This presentation provides a perspective on cloud computing's future and illustrates how the CIO can appropriately leverage cloud to become the hero again.’

Cloud is culmination of capability, developed over 50 years, with a new intriguing business model.  IBM is embracing Cloud Computing.  Today’s talk on why CIOs should embrace it as well.

Interesting, Mike Hill was IBM’s CIO in the ‘90s, driving internal e-business initiatives.  So, he should get the challenges of IT. 

The future: three co-existing delivery models: within the enterprise traditional IT & private cloud, public cloud and a hybrid cloud bridging private & public.

Use public cloud for cost efficiencies and to include public services in your SOA.

New challenges for CIO with cloud:

- scale & responsiveness (ensure)

- manage systems & security

- support audits & certifications

- manage proliferation of systems & applications

- integrate cloud and enterprise services

- provide a range of services

- provide a range of technical support

How to manage these challenges:

- develop cloud architecture and plan

- implement security, audit, and systems management technologies

- use a SOA

- demonstrate ROI

Mike Hill just mentioned the metric I missed from Kristof, 60-80%.  If return is not 60-80%, doing the wrong thing.  I’d have to guess that return there is cost savings, but I’m not certain.  Perhaps someone from IBM reading this can add a comment with some detail.  Mike just came back to this point, it is about % of cost taken out of running environment (related to targeted workloads of course).

Managing security, audit & compliance, and systems requires technologies for the cloud.  IBM is working on building the “right kinds of systems management software” to do this.  Mike cautions, as you are looking at public providers, be sure stringent enterprise requirements re: identity, data privacy, auditing, security & more are accounted for.

Now, Mike is “making an advertisement for SOA”.  “If you aren’t down that path, take a hard look again”.  Need to look at processes and services.

Mike is talking about KPIs, metrics, measurement and scorecards for IT.  Need to take cost out of IT and associated cost out of business.  I like the financial discipline aspects advocated by cloud vendors.  Mike says, if you don’t capture the cost taken out, you’ll never get to reclaim it to invest in new capability.

So, cloud could help IT keep a flat budget overall, but have more $$, higher %, to spend on new business capability delivery.  This, as I recall from my time in IT management, is the holy grail. 

IBM went through near-death experience not embracing a new compute model – client server – and reminds audience to avoid a near death experience.

Liveblogging: Cloud Computing Expo #8 Kristof Kloeckner (IBM)

Kristof Kloeckner, IBM CTO, Enterprise Initiatives and Vice President Cloud Computing Platforms on Cloud Computing and the Enterprise:

“This presentation will draw on IBM’s experience working with customers and operating 13 cloud centers world wide to review the conditions under which cloud computing can deliver its promise of flexibility and cost savings in the delivery of IT services to the enterprise.

Kristof Kloeckner will discuss the importance of dynamic infrastructures and service management for both public and private clouds, cloud service life cycles and integration between public clouds and enterprise services. He will also talk about standards required for interoperability between clouds and application scenarios that demonstrate the use of cloud computing in an enterprise context for a smarter planet.”

Cloud Computing Delivery Models: Public, Private and everything in between.  I’ve reported on this slide previously:

“From consumer perspective, Public  clouds offer standardization, capital preservation, flexibility and time to deploy.  Private clouds offer customization, efficiency, availability, resiliency, security and privacy.  Choice will depend on organization, culture, and governance.”

Cloud-o-nomics:

(1) virtualization + energy efficiency + standardization + automation (service management) = reduced cost

(2) agility + business & IT alignment + service flexibility + industry standards = optimized business

- allows you to optimize new investments for direct business benefits.

Kristof agrees with “some analysts” that cloud computing is related to, and benefits from, SOA.

5 Step Plan – IT Transformation Roadmap --- not a “rapture into the clouds for IT geeks”

1. Simplified | Consolidate – reduce infrastructure complexity, freeing up resources, improve business resilience, reduce TCO

2. Shared | Virtualize – remove physical boundaries between resources, increase hardware utilizations, simplify deployments,

3. Dynamic | Automate - introduce standardize services, service management, metering & measurement, granular services

Three roles in respect to cloud computing: end-users & operators (service catalog, operational console), service provider (cloud services & cloud management platform) & service creation and deployment (service definition, publishing, fulfillment & configuration, reporting & analysis).  Items in ( ) are primary concerns and tools for that role.  (Yes,  I imagine you can map this to ITSM and the Tivoli and WebSphere product lines.  Oh, and Rational, Kristof just mentioned that for modeling a service deployment topology.  (Krutchen 4+1 Architecture and all))

Customer discussions often start with “help me virtualize my workloads”.  Next, “can we move this into a shared, trusted, private cloud”. Finally, “what is possible to be deployed into public cloud”.

Given multi-cloud environment, need to ask about how services and data flow across clouds.  Need to understand security – identity & trust.  Also, connectivity or a (wait for it) “cloud broker”.

Ideally, information and services will flow freely across clouds, in secure and standardized manner.  This leads Kristof to the Open Cloud Manifesto.  He has the principles listed.  IBM is supporter, describes it as a first pass.  As you may have read, not everyone supports the Cloud Manifesto.

In analyzing workloads for cloud or not, need to:

- examine for risk: database, transaction processing, ERP

- test for standardization: web infrastructure, collaborative infrastructure, development & test, high performance computing 

- explore new workloads: high volume, low cost analytics, collaborative business networks, industry scale “smart” applications

4. Pick cloud model – private, public, or hybrid

5. Implementation

Ugh! Missed a good metric.  Kristof spoke of setting your cost savings metric and then measuring your cloud projects against it.  If project doesn’t meet metric, move along.  But, I missed the #. 

Now, IBM Cloud Computing Services Offerings…  Seems like “bursty workloads” are a target. 

Next, the obligatory global map showing cloud computing center locations.  Everyone had one yesterday as well. 

“start now with virtualization”

“modularity and standards are key”

Liveblogging: Cloud Computing Expo #7 David Bressler Cloud Computing & Enterprise Application Architecture

David Bressler, SOA & Cloud Evangelist for Progress Actional is up now, talking about the Impact of Cloud Computing on Enterprise Application Architecture:

“Cloud Computing will change enterprise consumption of IT, but challenges around data management, no longer “owned” by the individual application, pose a threat. These include: how to keep data accurate and in the right hands; how to add new sources; how to provide contextual information; and how to successfully drive this all into a business-users hands. In this session David Bressler, SOA Evangelist, Progress Software, will outline the “new enterprise app” vision where data relationships and their impact to the business process matter, and define best practices to ensure “safe” Cloud Computing that drives tangible business IT improvements.”

David promises to be entertaining, and since it’s now 7:00pm, I’m counting on it.

Punch line, turning commodities into utilities.

Cloud computing, culture of enterprise integration, will take a shot at defining cloud computing, then talk about some best practices.

“Integration is hard, by hard, I mean expensive”.  Integration is more than getting things to talk, it’s making things work properly.  Integration is for infrastructure and data.  People understand infrastructure and need to configure.  People don’t understand data integration and why it’s so hard.

Data integration is hard. Infrastructure integration, not so hard. 

Data integration isn’t just hard, but it’s hard to explain why it’s hard, because it’s invisible.  It’s hard to quantify.

With cloud computing, if we can make the easy stuff invisible, hopefully we can spend more time on the hard stuff, and get better at it.  Put more discipline around the things that are hard.

If you are in an enterprise and want to deploy cloud computing, or soa, need to build strategy and teach people why that strategy is important.

The whole way we buy software is wrong: all the upfront evaluation, purchases, delivery, configuration etc.  Today, all you need to do is get a login.  Spend $10/month.  You have the freedom to make mistakes, recover financially (the $10), and quickly change directions (get another login). 

Data integration is hard.  Costs don’t match benefits.

How to make it easier and deliver benefits more quickly?  “Cloud computing is a way to make integration easier and deliver benefits more quickly”.

Turn a commodity into a utility.  Not everything is going to be a commodity.  David calls out Xignite statement, don’t put it in the cloud if it needs to be secure.

Turn integration into a utility.  Move work that is necessary but not differentiating.

Commodities according to David: most enterprise software and email.

Messaging, not a commodity.  [Of course David is from Progress, which he readily admits]  His point is, he wants all his messages in one place.  Not, go to Facebook, Twitter, Salesforce etc.

David calls for building of platforms (me too).  Let’s spend [IT] time with data models, interfaces, processes  on a dynamic infrastructure [that met corporate requirements] & simply existed as needed?

This vision, achieving this vision, results in elevated IT relevance.  Computing to solve problems is really interested.

Why can’t we do “cool in the enterprise”.  Where cool is at the end-user & application, not in the server room.  Need to spend our time better.

Best practices:

- Mediation.  A secret weapon.  Powerful way to protect yourself from change.  Can also track who is doing what to whom.  He just used the “G-word” governance.

- Service Level Management. Don’t even start with “my piece is working fine”.  Need to abstract SLA up high enough so they make sense to the customer.  Not server up time, but application up time.

- Security. It’s not (only) what you think it needs to be.  Security needs to move up to the application level.

- Make mistakes.  Not 8 year beta, but doesn’t have to be perfect.  IT needs to focus on products, not projects.  Projects end.  Products have long term ownership and management.  

- Build a culture of collaboration.

- Map out a strategy. Reward innovation. Reward relevance.  Teach people why they are doing something. Let them choose how (within corporate guidelines)

David mentioned a related paper released by Progress today.  If I had wifi right now, I’d search for it. Thanks to Emily Estep for supplying the link.  [link added on 4.1.2009]

Liveblogging: Cloud Computing Expo #6, The Cloud, the Rich Client & the Buckeye State

Introduced as “one of our more unusual sessions”.  Session abstract:

“State government a hotbed of technology innovation - huh? But that’s exactly right. The Ohio Dept of Transportation has harnessed extensive infrastructure (thousands of sensors, historical & police data, etc.) and cloud services to create a rich solution. Through a series of demos, this session will showcase for delegates how the Buckeye State has created a rich user experiences for residents, visitors and internal staff conducting detailed traffic analysis. In doing so, we will see how the IT team looked beyond just using cloud services, to combining with what’s on premise, and then becoming a supplier of these services to others. We will also share follow-up resources that attendees can review for further reference and follow-up, including access to code libraries, code snippets from the discussion. Join the two presenters and see for yourself the combination of on-premise & services in action.”

Session speakers are Brian H. Prince, an Architect Evangelist with Microsoft and Steve Milroy,a solution architect with OnTerra Systems.  The ODOT architect is home in Ohio, economy & travel restrictions.  Sound familiar?

The session is doing a demo of buckeye traffic, a publicly available site.  I just went out to check out road closures.  The data for the road conditions – icy, wet – are provided by sensors in the road.  There’s also webcam integration and a bunch of interesting data.  The public layers are only part of the solution, there’s functionality for ODOT, first responders and more.

Now, the cloud part.  The mapping application is Virtual Earth.  Upon a filter click request, the ODOT central repositories are accessed.  These repositories and corresponding web services are on-premise.  The traffic and weather sensors feed a variety of applications for ODOT.  There is also robotic and manual data collection. 

There is a strong customer community.  Demand for the application varies by external events, such as weather conditions.  This is one reason why ODOT moved from an on-premise mapping solution to Virtual Earth. 

Picking on-premise vs. cloud

- only write the code that only you can write!

- harness investments in on-premises infrastructure – can’t just throw out what you have in place

- development cost and time – ODOT reduced both cost & time

- infrastructure costs – during spikes (floodgate events), ODOT is no longer serving mapping images out, just responding to data request; scale is now off-premise

- bandwidth usage

Going Forward

- Increase enterprise innovation – this project kick started ODOT’s innovation process/projects

- Make IT part of the Conversation – IT moved to adult table

This solution is used by citizens, agencies and vendors.  Both the mapping application and the underlying data.

“The magic behind SOA that cloud makes it easier to execute is that I can now share information and applications with anyone that is interested.” –Brian Prince

In closing, advise to consider moving applications that are “core but not strategic” to the cloud.

Liveblogging: Cloud Computing Expo #5 Forbes.com in the Cloud

Jeff Bauer from Forbes.com starts by describing his business challenge: Make real-time quotes available the website, do it fast & cheap and the history of real-time quote availability and how the owners, availability and price of real-time data has changed overtime. 

Decided to use BATS exchange for real-time data, gave them access to streaming data, but the data was in a raw format.  Forbes didn’t have the infrastructure to transform the raw data into response to real-time stock quote requests.  Forbes is a publishing company, not a technology provider.

Forbes.com decided to partner with Xignite.  With that segue, Joel York of Xignite is up to explain why cloud computing made sense for this particular application.  Xignite is a cloud services platform for “new generation” financials applications.

Xignite offers market data as a service, in the cloud.  Xignite has been doing this for 5-8 years. 

Xignite leverages Amazon EC2 in the Forbes solution.  The biggest driver of this choice is the huge variation in compute needs that correspond to when the markets are open.  Owning infrastructure would have it nearly (comparatively) unused on the weekends and during the week when markets aren’t up.

The heart of the solution is a ticker plant.  Xignite shared  that they are a Microsoft development shop. 

Here’s a press release on the Forbes solution. 

Interesting point in the Q&A, by being a cloud services provider -- data services -- rather than a cloud application provider (SaaS), it is easier to inject customization for a customer.  Essentially, you can use the same services in a variety of compositions.  [Yes, sounds like SOA].  A SaaS application designed for multi-tenants does not welcome customization, because that breaks the (economy of scale) business model of the provider.  Joel comments that client customization needs drove creation of Force.com.

Now, the Q&A is talking about security and how the Forbes.com real-time quote application was low risk, because it’s not Forbes’ data going out to the cloud (to Amazon).  It is third party data that is widely available. 

Other questions asked, what about maintaining SLAs?  Xignite uses Rightscale.  From the answer, I can’t tell if the SLA is between Xignite & Amazon, Xignite & Forbes or both.

Always good to see/hear a case study.  Another coming next.