Forgoing the hyperbole of cloud computing predictions – sensational outages to a cloud-in-every-pocket – I want to start 2010 discussing the enduring aspects of cloud computing on enterprise business-technology.  Regardless of the final manifestation of cloud computing, and the tally of deployments, successes and failures, I believe cloud computing will influence the expectations and practice of enterprise business-technology throughout the decade.

I have identified five enduring aspects from a practitioner perspective.  Certainly, there are enduring aspects on the provider side as well, such as advances from Infrastructure 2.0 and disruptions created by new economic and pricing models.  However, I will leave that list for provider-side specialists. 

The first three enduring aspects focus on the expectations from business-technology organizations. 

1. Resource Optimization – Cloud computing has raised Executive awareness to the disproportion of installed versus utilized computing capacity, along with the requisite expenses of space, power, software licenses and support personnel. 

If they have not already, Executives will mandate infrastructure ecology initiatives, starting with the consolidation and pooling of compute and data resources, and progressing to software execution efficiency. 

 more >>

Posted by brenda michelson at 5:59 pm in Blog, economics, elasticity & scale, fundamentals, infrastructure 2.0, performance & reliability, platform, pundit positions, software architecture | Permalink | Comments(4)
| Trackback URL

Via Joe McKendrick, I came across Charlie Bess’ post listing some metric suggestions for cloud computing adopters.  Bess’ list:

  • Provisioning time (down, leading)
  • Service reuse  (up, lagging)
  • Utilization (hardware, storage…)  (up, lagging)
  • Uptime/hour  (down, lagging)
  • In house personnel dedicated to operations support (down, lagging)
  • Business value generated per effort hour (up, lagging)
  • Business value generated per watt consumed (up, lagging)
  • % of IT budget dedicated to fixed costs and maintenance  (down, lagging)

In reading the list, note the “items in parenthesis are first "the good trend" and if it was a leading or lagging indicator.”

This will be an important topic in 2010, and Bess’ list is a great starting point.  I especially like the business value focus, and the recognition that cloud computing does incur in-house personnel costs.  As Bess mentions, there needs to be more leading indicators, otherwise organizations might be inclined to stay with known costs, rather than introduce unknown expense, and of course, risks.

This is definitely a stream I’ll be following over 2010, the real-costs, value and risks.  And how the value proposition changes with the element of time.

Posted by brenda michelson at 11:26 am in adoption, Cloud Watch, economics | Permalink | Comments(0)
| Trackback URL

The good folks at SOASTA have secured $10 million in Series C Funding:

SOASTA, the leader in cloud testing, today announced the successful close of $10 million in Series C funding as it prepares its global expansion. The round was led by UV Partners, and included participation from all existing investors: Canaan Partners, Formative Ventures and The Entrepreneurs’ Fund. The company’s growth plan includes the opening of offices in Europe, broadening of the reseller channel and increasing its technology roadmap with product enhancements to address the growing demand for SOASTA’s cloud-based testing service.

Today’s announcement follows other significant company milestones. SOASTA recently announced a new partnership with Computer Sciences Corporation (NYSE: CSC), a leading managed services provider, who has integrated CloudTest into its Trusted Cloud Services offering and its testing and development methodology. The company completed large web-scale tests with Best Buy, Hallmark, Leapfrog, M-Dot Networks, MySpace, Schlumberger, SAP and Zappos.com, and also announced open source support by offering JMeter users the ability to run their scripts in the SOASTA Global Test Cloud.”

About CloudTest

“SOASTA CloudTest On-Demand is a full-service offering. Customers simply describe the web user business process, such as logging into an account, executing a transaction, or browsing content. SOASTA’s team of experienced performance engineers build the tests, provision the complete cloud environment, execute the tests, and work with customers to analyze, fix and tune a site’s performance. SOASTA CloudTest’s unique, real-time metrics and analytics of massive test results data gives customers the performance intelligence they need to pinpoint and fix issues as tests are being run — ensuring greater confidence in website reliability and performance.”

[Disclosure: SOASTA has done business with my firm, Elemental Links, in the past.]

Posted by brenda michelson at 2:14 pm in cloud offering, Cloud Watch, economics, PaaS | Permalink | Comments(0)
| Trackback URL

November 16th, 2009

SOA, Cloud & Value Acceleration

Given my longstanding background in SOA and my current work in Cloud Computing, I’m often asked how the two complement each other, particularly from the business benefit perspective.  Briefly, here are my thoughts and observations.

From the SOA Side

The immediate benefit of combining SOA and Cloud Computing is time.  Reaching out to the cloud for business or technology capabilities, allows SOA initiatives to compress time to value. 

In the longer term, the benefits include improved collaboration, customer satisfaction and business growth.  By offering SOA based business capabilities to the cloud, businesses can improve interactions with business partners and existing customers, and/or generate new revenue streams.

From the Cloud Side

The immediate benefit of cloud computing is financial.   more >>

Posted by brenda michelson at 4:06 pm in adoption, Blog, economics, enterprise architecture, services architecture | Permalink | Comments(0)
| Trackback URL

The Economist is hosting a cloud computing debate.  Their opening position is “the cloud can’t be entirely trusted”.  Parsing the sentence down to “entirely trusted”, taking a position of “complete disagreement” is unreasonable.  Semantics aside, the debate has some good contributions.

George Gilbert, of TechAlpha, offered a guest commentary on Friday.  Gilbert’s commentary speaks to the necessary industry transition for widespread cloud computing adoption, focusing on security issues, and relinquishing management control.  Gilbert then speaks to the economic implications for organizations adopting cloud computing, and the vendors currently supplying those organizations.

“Adding to the challenges of the industry transition, some vendors just don’t want it to happen—and none with more intransigence than Oracle. The implications for business models are cautionary.

In an on-demand world, customers will be able to buy and deploy capacity "just in time" instead of "just in case". This has potentially profound effects on both hardware infrastructure vendors and server software vendors such as IBM, Microsoft and Oracle. In essence, even when customers choose to own their hardware infrastructure and software in the future, there will be tremendous pressure on vendors to price in the same on-demand way that customers will consume it.

Today, infrastructure hardware and server software capacity is consumed in a chunky manner because the software licences are typically allocated and bound to a physical box in perpetuity. To compensate for a lack of flexibility in provisioning, it is not uncommon for customers to purchase three years of capacity upfront. In IBM’s published explanation of its cloud-friendly pricing for its database which enables the use of incremental capacity, it claims average database server utilisation in its customer base of 5-20%.

As one Fortune 100 CIO put it succinctly, "Buying minutes of capacity that can float across different physical machines fits the current economic constraints a lot better than buying perpetual capacity tied to a specific physical box." Software vendors are already adopting this pricing model on public clouds like Amazon Web Services.

The transition to a pricing model where customers are able to pay for smaller increments of capacity in smaller increments of time will be highly disruptive to current vendor business models. The change pushes vendors to move closer to a utility-like subscription pricing model. For software companies in particular, having upfront recognition of perpetual licences give way to subscriptions would have a material impact on recognised revenues and reported earnings. Even if vendors were able to bill one or two years upfront, the cash flow and reporting of non-GAAP earnings would still not make up the difference.

Is there any wonder why Oracle’s CEO, Larry Ellison, dismisses cloud computing as nothing more than "water vapour"?”

Posted by brenda michelson at 10:22 am in Cloud Watch, economics | Permalink | Comments(0)
| Trackback URL

Datacenter Knowledge published an interesting article today entitled Data Center Leasing: It’s All About the Megawatts.  The article discusses how electric power, or the lack of it, is changing the business of data center leasing:

“The growing importance of electric power is remaking the business of leasing data center space, with megawatts replacing square feet as the primary benchmark for real estate deals.

“Our business is all about leasing access to power,” said Michael Foust, the CEO of Digital Realty Trust, the largest data center developer landlord. “The square footage is almost secondary in some cases.”

The consumer perspective:

“If your company isn’t measuring whatever the widget is by kilowatts, you need to,” said Chris Crosby, VP of business development at Digital Realty. “Power is really the key element for measurement. The clients that understand things in these terms make the best decisions.

“If you talk about square footage in the data center, you’ll get confused in your buying decision,” Crosby added. “Getting an understanding of your kilowatts will give you a lot of insight into your costs.”

The article then describes a leasing decision scenario for Currenex:

“An example is Currenex, an electronic trading platform specializing in the foreign exchange markets. The company’s business was growing quickly, but its two third-party data center providers had limited capacity.

“We were operating in data centers that were out of power,” said Chad Parris, VP of technical operations at Currenex. “We’re not big enough to build our own facility, so it became tempting to buy cage or cabinet solutions (in colocation centers).”

Instead, Currenex performed a detailed analysis of the power required to execute its trades, and then used those numbers to convert daily volume of foreign exchange trades into kilowatts. This allowed the company to model its future power needs based on estimates of its future trading volume.

Currenex wound up leasing 2,500 square feet space in a Digital Realty data center in New Jersey, with an option for additional space as its operation expands. The wholesale data center model “was the cheapest way to buy a long-term position in kilowatts,” Parris said. “In year four, our cost per kilowatt a month is about a third of what it would have been.””

——

Why is this interesting?  After all, data center leasing doesn’t equate to cloud computing.  As the economy recovers, there will be greater demand for all types of energy, which will raise all energy prices, including electricity.  Understanding the kilowatt component of your workloads will enable better financial analysis on transaction run costs across computing environment destinations: traditional datacenter, outsourced datacenter, internally owned and managed clouds or third party owned and managed clouds. 

Additionally, if the cloud computing environment providers start breaking out kilowatts as a pricing factor, you will be prepared.

Posted by brenda michelson at 3:51 pm in billing & metering, Cloud Watch, economics | Permalink | Comments(0)
| Trackback URL