Moving to the Cloud

One of the main workloads that Microsoft’s Azure PaaS offers superior value for are elastic computing workloads. Especially workloads where consumption is not easily predicted. You will find Seasonal, Fiscal and Social workloads to be highly variable. Seasonal affects consumer scenarios like Christmas, back to school. Fiscal affects budget planning, Taxes. Social can be very unpredictable, you never know what may go viral. Some applications may also benefit from “hibernating”, our HR enrollment period for example happens for a few months in the year and then goes dark.  The consumption based model plays very well here and brings dramatic cost savings to the run state of  applications.

Another up and coming workload type would be mobile Services especially net new services may be easier to build out in the  cloud than on premise in the traditional way.

 

From an IT portfolio perspective there are two classes of workloads that will benefit from the cloud.

·        Existing –primarily Azure IaaS or Private Cloud IaaS
·        New

Existing workloads are ones that we run but are not making engineering investment in. These workloads represent about 65% of Microsoft’s workloads and  for most other companies 80%.  The key value proposition for the New workloads are varied … in principle IT follows the Leverage, Buy, Build approach for capabilities/workloads it needs to invest in so …

Leverage – larger companies are riddled with duplicity of applications that support the same capability. Before investing in buy or build IT needs to see if there is existing solutions already running in house and leverage that. Better time to  market, better consolidation of requirements, better cost efficiencies. The cloud play here is the “new” workload my offer investment to the existing capability so that it can justify moving to the cloud. First you need to know if you have good connectivity to the public cloud such that design part of your  eco-system in the public cloud will meet your requirements. If latency, scale, security or risk issues preclude public cloud adoption then you will have to  invest in the traditional model of development and deploy to Private Cloud IaaS.

Buy – more and more we are looking towards application or solution vendors to deliver capabilities vs. developing in-house. These vendors are the primary target for PaaS. Not IT. IT’s real job isn’t developing applications. It’s evaluating, planning, deploying, running and becoming a Partner with business on strategic use of Technology.

Build – Most IT shops don’t have much capacity to build applications for their companies. Typically large enterprises have some capacity for this but small to medium companies don’t. The class of applications that IT should build are ones that provide a strategic business/technical advantage. One that defines the company.  Like trading analysis applications for stock brokers, or streaming clients for media companies. In this regard IT is a software company akin to the target for PaaS (application developers or SaaS vendors). The strategy should be to build new capabilities or new subsystems in the cloud if connectivity, security and capacity requirements can be met. If not they the application should be designed so at it doesn’t preclude moving to the cloud. Architectural principles like employing a scale out design vs. scale up, accounting for longer latency between each layer (presentation, service, data) and improved resilience (like automating connecting to the data tier vs. throwing exceptions).

Note: Most IT companies have a small %of apps they can change during a year and of that a smaller that they can redesign. Typically its 10-30% change and of that 10% that can be redesigned (1-3%).

Bottom line: moving workloads to the cloud needs to focus on application and software as a service providers. It’s a channel strategy who in turn will deliver this value to IT.