Two basic questions
- What is cloud computing, and how does it potentially play a role in higher education?
- Virtualization should be enough for an ecosystem (universities together or a collegiate of institutions)... or is it?
Here are a few things about academic computing environment.
Spiky – seasonal traffic loads are multiples of daily load
Traffic on our services in higher education follow predicable patterns. We have application spikes, registration spikes, testing and other scheduled activities that drive loads in our environment like examination results to name one. Once in a while, a special event will create an unusual spike in data volume and not just concurrency. This is more and more true as we go towards an ecosystem of using video and multimedia. However, in general, we can predict our ‘spikyness’, the dates and time... but managing it has always been a situation.
Slow provisioning sets the stage for inefficient use of resources
Even with virtualization, we find that it often takes days to get a new service provisioned in a data center. How quickly can we enable a new campus to run live on an ERP (for instance) to execute all the business processes, even if they are very similar to other campuses. We may have the tools, like fully stocked SAN and storage virtualization. Unfortunately, a fully deployed service requires much more than just the bare VM instance. There are many other aspects of deploying a service which often require time consuming manual configuration. The problem is that each server has more than the initial fixed cost. There are power and cooling costs, support costs, operational costs etc. So here are the challenges we hear from our organizations:
- How can we respond more quickly to need?
- How can we use resources more efficiently?
Cloud requirements from an academic perspective
In order to meet those challenges, we need to address several significant obstacles:
- Critical data needs to be secure
- Provisioning needs to include network and application configuration
- Application instances need to be fully portable within nodes in a cloud cluster
- Application instances need to be location agnostic
- Cloud needs to be technology agnostic – cloud manager, hypervisor, application stack, application configuration, network configuration, storage provisioning
Secure handling of data is critical
Like financial industry, secure handling of data is a crucial component of any cloud strategy. When it comes to security, cloud computing gets interesting. The problem is of course compounded when the cloud extends beyond the internal campus network. The problem with a private cloud inside the firewall is that if the resources are being used by a broad user base across the campus, any one virtualized instance can request access for the outside world. An infrastructure as a service (IaaS) approach can create situations where one virtualized application service could be handling secure data while another virtual instance on the same cloud could be compromised. Does this pose a security risk to private data? Conceivably someone hacking a public instance could hop to other instances on the cloud – private or not!
For higher education, the risks and the rewards are high. There will be breaches and corresponding responses. Is our data more secure on our own servers? Risk is a daily balance and there is no such thing as a risk free environment. The question you need to answer is: as an organization, can we do better on our own or does leverage a larger organization’s information security resources make us safer, even if you are part of a richer target. This is not the first time we’ve faced this question in the evolution of our species nor will it be the last. If you choose to go with the larger institutional strategy (maybe executing around a collegiate), it is important that you keep up with software updates and the latest trends in applications and technology. Otherwise you risk the worst of both worlds: a juicy target that doesn’t keep up their defenses.
Reference: Dennis Anderson, Ph.D and Dr. Peter Morales (Is Cloud Computing Ready for Higher Education)