Cloud Expo NYC (and of course...Big Data)

0

I recently attended the Cloud Expo in New York City where big data was a key topic. Keeping with the various topics addressed at the event, here are my thoughts as they relate to the Cloud Expo. Warning: if you are searching for strong conclusions or straightforward advice, you’ll need to look elsewhere! 

  • What is more “stratospherically” hyped? Most would agree that big data hype has eclipsed cloud hype, but if anyone can come up with an acronym that combines big data, cloud, social media and mobile, then we’ll have the ultimate hype machine.  
  • Big data and cloud – a match made in “heaven”? I’m conflicted about this combination. Certainly elastic compute power, rapid provisioning, measured services, etc., are a perfect complement for managing big data. But most people think public cloud and big data, and therein lies the problem. It’s highly unlikely that all of your big data is in the public cloud, and if so, it probably isn’t in the same cloud infrastructure that you will leverage for analytics, so that means data movement. Although there are various forms of high-speed file transfer technology, change detection methods that can isolate updates, federated queries, pre-processing methods that limit data movement to intermediate results, etc....there is still a physics problem when it comes to moving data. I think we’ll find that private cloud deployments provide a better fit for projects involving massive data volumes since there will be less friction involved with moving data.
  • Speaking of private clouds, how many organizations can truly “cloud seed”? Although many organizations are using basic cloud principles like virtualization, how many organizations are really capable of building and managing a private cloud that is not just a glorified virtualized compute infrastructure? I personally believe it is a relatively small number, which means that organizations will turn to cloud management vendors – even so, the work required to spin up an application on a public cloud is trivial compared to what it will take to use RackSpace, VMWare, HP, etc., to deploy a true enterprise-level, private cloud. And then you have to make a bet on the right cloud infrastructure – including selecting the right API / service standard to use – OpenStack, CloudStack, etc. 
  • What type of cloud is it? “cumulus, stratus, cirrus or nimbus”? Companies that I speak to are clearly confused about what cloud deployment approach to use – should they leverage a public cloud or a private cloud? What role should a hybrid cloud play, and if it is private, is it virtual private cloud in the public cloud, or a private on-premise cloud? And, what is a community cloud anyway? There are many considerations – security, manageability, financial accounting, etc., and although one of these may trump other considerations, the application use case should be a key factor.  The industry should work towards defining a set of analytical patterns that map to a preferred deployment environment. 
  • Avoid a “cloud appendage” at all costs. It’s beneficial to ensure that your cloud strategy will not result in yet another silo. Don’t fall into a false sense of security just because many cloud deployments are web based – a SOA-based interface doesn’t  eliminate integration issues. And, it’s not just about integration, organizations should extend their governance, management and policy approach to their cloud initiatives. 
  • Don’t accept a blanket “cloud is cheaper forecast”! First of all it’s not all about cost, but even if it is, a public cloud option may not always be the most cost effective. Look closely at your application profile – is it a long term or short term application? Does it have consistent usage or does the usage occur in spikes? Spiky usage is good fit for cloud – you can scale up for peak:  “buy the base, burst the spike”. But that assumes that you can design your infrastructure so that you can effectively burst, which is easier to do if you have a vertically integrated infrastructure. 
  • Consumer experience is driving “heightened” expectations. Don’t underestimate the expectations of your users – especially the Millennials. They are accustomed to the instantaneous service – whether it’s hosted email, social sources such as FaceBook, Twitter, etc. They may not think about whether the service is in the cloud, they may not understand the cloud, but they will expect the simplicity of the those services. After all, “there’s an app for that”. 
  • An “abstract cloud” should allow the developer to focus. You’ll need a complete team that can manage virtualization / Hypervisor support, storage, compute, as well as service management, but the key is to provide a level of abstraction to your developers. Allow them to focuson operational or analytical services that will provide business differentiation vs. having them fight the infrastructure. It should be about making it easy for them – spinning up an environment, whether it is dev, test or production, providing them management support, etc. 
  • New approaches are necessary for “cloud control”. It’s not always best to take legacy applications and move them to the cloud. As with other new technologies, the cloud may be more appropriate for new workloads. It’s important to think differently, designing for the cloud has different considerations - just taking legacy and throwing it into the cloud may create issues. It’s best to design and build specifically for the cloud. 
  • When it comes to Software as a Service (SaaS) it’s not just “pay by the drink” (ok, I ran out of cloud phrases!). There is a lot written about how software is licensed or procured – the move from perpetual licenses to a usage based model typically provides more flexibility. But licensing is not the only change… the cloud will alter your interactions and your influence – if you rely on a public cloud infrastructure, you no longer interact with the hardware manufacturer, your negotiations and your agreements fundamentally change – you rely on the cloud vendor for everything – service levels, hardware support, etc.

And to think that I addressed these topics without mentioning salesforce.com!

For more on Cloud Expo: http://twitter.com/CloudExpo

Share

About Author

Mark Troester

IT / CIO Thought Leader & Strategist

Mark Troester is the IT / CIO Thought Leader & Strategist for SAS. He oversees the company’s market strategy efforts for information management and for the overall CIO and IT vision. He began his career in IT and has worked in product management and product marketing for a number of Silicon Valley start-ups and established software companies. Twitter @mtroester

Leave A Reply

Back to Top