How many connected devices do you own?By 2020 we will have to support about 24 billion devices. That’s only 7 short years away!Can your current data center support a 3x increase? What are your plans to support the increase?http://gigaom.com/cloud/internet-of-things-will-have-24-billion-devices-by-2020/
The biggest cloud drivers are:Scalability – supporting that 3x increase by 2020Agility – current data center thinking in not very flexible. We allocate resources based on projects. We ask our architects how much CPU, memory, disk space a project requires and build systems from there. What about point #1? If an application needs to scale, can it?Cost – as with most things, cost saving is always good. One point here is, we need to utilize what we have more efficiently. Virtualization took us from 5-15% utilization to about 30-50%. Think about that, data center resources sitting idle for 50-70% of the time. We have to get better at this.Remember – Can your data center support a 3x increase?
What our future looks like. As we move toward our 2020 goal of supporting 3x the workloads of today, we will use a mix of public and private cloud to support the increasing demand. Or what we call Hybrid Cloud.
http://csrc.nist.gov/publications/drafts/800-145/Draft-SP-800-145_cloud-definition.pdfPretty much any IT survey in the last 2 years puts cloud and virtualization among the top goals for most companies.So what is cloud computing? Seems like every company says that they have a cloud today, but do they? Are they just cloud-washing?NIST offers a broad definition covering all aspects of cloud computing.More detail on the following pages.
From the top done.Deployment modelsService models5 Essential Characteristics
What does a cloud look like? What characteristics does it need?On-demand self-service – Self-service to me takes on two characteristics. The first is a for individual consumers of the cloud to have portal like access. The second is to allow programmatic access. An API. This is for build and automation tools, orchestration engines, monitoring frameworks. Network access – People have to able to access what we build on these cloudsResource pooling – combine our resources. No more silo’s or equipment. Let others shareElasticity – the ability to change or be flexibleMeasured service – charge back or show back
AWS: Amazon Web ServicesGCE: Google Compute Engine
Approvals and deployments depend on humans and therefore have the potential for delays. Deployment can take days, weeks, or even months.
The cloud itself, not the IT department, checks to see if the request is within policies. Because humans are less involved there is no potential for delay.
How are we going to get there?Server virtualizationDistributed virtualizationPrivate Cloud – EucalyptusHybrid Cloud – Eucalyptus and Amazon AWS – this is were we can share the load across the on-premise Eucalyptus cloud and the public Amazon AWS cloud. This gives you the maximum flexibility.Public Cloud- Amazon AWS – moving as much of your workloads into the public cloud.Link to Gartner research:http://www.internap.com/wp-content/uploads/Virtualization-to-the-Cloud-Internap_vol2_issue2.pdf
Bursty workloads – seasonal spikesTransient apps – QABig data – move from data-warehousing analyzing in real timeGaming – could be spikey
How are we going to get there?Server virtualizationDistributed virtualizationPrivate Cloud – EucalyptusHybrid Cloud – Eucalyptus and Amazon AWS – this is were we can share the load across the on-premise Eucalyptus cloud and the public Amazon AWS cloud. This gives you the maximum flexibility.Public Cloud- Amazon AWS – moving as much of your workloads into the public cloud.Link to Gartner research:http://www.internap.com/wp-content/uploads/Virtualization-to-the-Cloud-Internap_vol2_issue2.pdf