London Bridge is Falling Down
The medieval London Bridge suffered from overcommit, now facing cloud computing. Resource pooling offers economies of scale, but not when resources are needed at once. Like the bridge, it can support houses or massive traffic, just not both.
Everyone is familiar with the traditional nursery rhyme, “London Bridge is Falling Down.” However, few know that it traces its roots back to a factual wonder of the medieval world.
In 1209 a massive stone bridge was opened over the river Thames. Quite different than the modern London Bridge we know today, this colossal structure was an engineering marvel of its day and included a chapel at the apex of the bridge.
It didn’t take long for people to realize the potential of this new prime real estate and by the late 1200’s the bridge was completely lined with multi-story structures straddling each side of the bridge (some extending out over the water). The bottom floor was comprised of businesses while the upper floors became the most desirable housing in medieval London.
Somehow this sounds oddly familiar: Taking a piece of shared infrastructure and placing business and consumer services upon it, enabling a new way of doing business and living our lives.
The problem, as you can likely tell by the sole musical remnant of this once massive structure, is that it quite literally crumbled.
The original London Bridge struggled from a problem facing the cloud computing model… overcommit. Resource pooling provides fantastic economies of scale, but what happens when everyone needs the resources at the same time? The bridge could support the houses or a massive amount of traffic, just not both.
As an example, Amazon Web Services reserves memory for each instance, however CPU may be reserved or shared based on the instance type. VMware vCloud has a configurable Allocation Model allowing service providers to control the percentage of resource guarantee.
When resources are reserved, rejections happen cleanly when provisioning instances. In the overcommit model there is a risk of unpredictable failures at the application layer during a ‘perfect storm’ of resource usage. In the physical world, too many heavy carts crossing the bridge at once results in structural failure; in the virtual world overcommitment could mean data loss, data corruption or lack of availability.
So this begs the question: Are you running on a “London Cloud,” my fair lady? Make sure you know what resource allocation model your cloud uses.
[Ed. note: Trend Micro would like to know what you think about this. We enthusiastically invite your comments and we will read every one of them. For very detailed information about Trend Micro and Security Built for Enterprise Virtualization and Cloud Environments, please visit our website: http://bit.ly/dEmlhv ]