Last night was a particularly uncomfortable commute home on the subway. After watching three trains pass by from overcrowding, I finally managed to squeeze in behind a few other frustrated commuters on a humid train ride home.
This experience got me thinking. How does the city manage capacity on the subway? Clearly in this city there seems to be a gap between the amount of people who ride public transit each day and the amount of people the subway cars seem to be able to handle. But how exactly do you define capacity? More importantly, how do you define an ideal level of capacity and control it?
Is the ideal capacity of the subway simply the maximum amount of poor souls you can squeeze in to each train? Are the worker bees nothing more than clowns being squeezed in to an old VW bug? How do you balance the real needs of all the users, with the available resources of the system? Are the requirements of all users the same, or should special consideration be taken for the elderly or children? How do I avoid having to sit next to “that guy”?
These same questions are being asked in the datacenter.
What is the best way to manage multiple users sharing a central cluster of servers? How do you balance the usage of different levels of user, or different tier of applications?
When virtualization emerged, the idea of squeezing more and more users in to less and less resources seems the most natural way to cut costs. Today as the technology has matured users now find that was that an unrealistic expectation. In fact without adequate planning many virtualization projects face the opposite problem; they are simply not squeezing quite enough users or VMs in to the available virtual resources to begin with. Capacity is simply not being used appropriately.
Comprehensive planning tools can help administrators to better understand the available capacity they have. Looking back at historical usage information in both the old physical world and today’s new virtual world helps build a clear understanding on how best to use new server capacity. These tools can help to analyze and report on usage and available resources to get an understanding of best practices of where to slot VMs to manage resource capacity.
As virtual infrastructure continues to mature towards becoming a resource cloud, managing optimum capacity usage can also be done in another way. Up front. By building policy in to the initial requests for virtual resources, organization can now manage their capacity in an even better way, by allocating just enough resources to each user at the time of request.
Seems to all make a lot of sense to me, to build an elastic infrastructure, usage should be dictated by policy up front. Once things are up and running, thorough reporting and analysis allow you to catch errors in allocation of capacity after the fact so that you can make corrections.
If only the subway management thought of this…
Disclaimer: As with everything else at Cool Solutions, this content is definitely not supported by Novell (so don't even think of calling Support if you try something and it blows up).
It was contributed by a community member and is published "as is." It seems to have worked for at least one person, and might work for you. But please be sure to test, test, test before you do anything drastic with it.