You are on page 1of 4

Every time I talk virtualization and cloud security, I am surprised

by the number of people who do not follow the basics. The basics
have been around since 2004. The funny thing is, I also see the
basics misquoted all the time. People try to interject their language
into something already simple, yet by doing so, they limit the
overall idea behind the basics. What are these basics I am talking
about? The recent Virtualization and Cloud Security Podcast went
back to basics to cover the basics in detail. You can find the podcast
on iTunes and Talkshoe. Here is a rundown of getting back to
basics.

Virtualization and cloud security basics started when a little-known


engineer at Digital Equipment Corporation was talking to his
customers about the need to place a firewall in front of the
management constructs within each hypervisor. Yes, that engineer
was myself, aka Texiwill. Things have since expanded to many
different aspects of security. Let us get back to basics together:
 Improve monitoring so that you know who did what, when, where,
and how. In essence, you need a security operations center that
looks at activity within your virtual environment. This operations
center will help with security as well as operational issues. The
operations center should track users as well as actions to users.
Lastly, the operations center should alert when disallowed things
are done, such as actions taking place as the administrator and root
users, or when new virtual switches come into being, or
promiscuous mode is allowed within those virtual switch
constructs. Improving monitoring and alerting is the most basic
thing anyone can do to improve the overall security of a virtual or
cloud environment. Get a handle on what is happening in as much
detail as possible.
 Never use the administrator or root users by your management
tools. When you join a new management tool to your management
constructs, give it a unique username. This way, you will be able to
distinguish users from tools and to tell who did what. Administrator
and root users should only be used in break-glass situations by
humans, not machines.
 Protect the management network of any hypervisor cluster, usually
by placing the management network behind a firewall and enabling
multifactor authentication. The management network contains
anything that directly speaks with the hypervisor management
construct or the management tool itself. In the world of VMware,
this is anything that talks directly to ESXi’s management construct
and directly to vCenter. This includes the VMware management
tools found within vRealize Suite and NSX. For the world of KVM,
protect anything that talks to Dom0 and oVirt. In the world of
Hyper-V, it would be anything talking to the Hyper-V server and
System Center Virtual Machine Manager, and for Xen it would be
Dom0 and XenConsole. In the world of cloud, this is the cloud
management portal. This basic rule is just to protect the
management constructs; it is not to protect the workloads.
 Protect your storage (VSAN, VSA, iSCSI, NFS, and Fibre
Channel), live migration (or vMotion), fault tolerance, and backup
networks. Many of these networks transfer data in clear text and
now travel over entire data-center networks instead of private
networks.
 If you use VLANs as a segmentation tool, please be aware that
VLANs’ fundamental purpose is to improve performance of
switches, not to improve security. Switches are still under attack, so
you need compensating controls to ensure that the configuration of
your switch is not changed and to ensure that nothing plugs into a
switch that shouldn’t be there. For physical switches, that is usually
the case, but for virtual switches, that is not the case. Improve your
monitoring for network security into the virtual switch realm.
 Change your mindset. Stop thinking about virtualization as a single
node; think of it as a cluster of nodes with interconnects between
the nodes. Even the smallest of the small have minimally a two- to
three-node cluster, while the largest can have hundreds, if not
thousands, of nodes. Virtualization security transcends the single
node, single system concept.
 Take the time to get educated and to educate your security and
compliance folks. We have said in the past that the virtualization
and cloud administrator needs to lead the charge, to help educate
folks. This is far easier today than it has ever been
with vBrownBag, The Great vSwitch Debate, YouTube, my own
books, and other web-based training resources.
None of these actions cost any money to implement; they do,
however, take time. Some even take some planning. Those plans
may show you need more hardware or software to implement these
items adequately. This is not a debate about using VLANs or not; it
is about considering things first without technologies, then adding
the technologies you need to secure your environment. The simplest
way to go back to basics is with monitoring. Use monitoring to
guide you down the other steps, to improve your architecture, and
ultimately to secure your environment.
Have a listen to the podcast and let us know your thoughts!

Edward L. Haletky, aka Texiwill, is an analyst, author, architect, technologist, and out-of-the-box thinker.
As an analyst, Edward looks at all things IoT, big data, cloud, security, and DevOps. As an author, he has
written about virtualization and security. As an architect, Edward creates peer-reviewed reference
architectures for hybrid cloud, cloud-native applications, and many other aspects of the modern business.
As a technologist, Edward creates code prototypes for parts of those architectures. Edward is solving
today’s problems in an implementable fashion.

You might also like