Storage Provisioning and Networking - Deploying and Managing a Cloud Infrastructure - page 254

However, even though a tenant may have on-demand access to security controls such
as virtual firewalls, authentication services, and logging, things could change swiftly if the
underlying implementation is patched or updated (by the service provider).
Moreover, as you reprovision the VM in an updated infrastructure, the security configu-
ration data might become operationally incorrect. To curtail such problems, configuration
management and version control systems might also need updating.
The biggest concern of all is the unintended interactions and information transfer when
on-demand security controls are integrated with a customer application. What is possible
is that if machine IDs and IP addresses are recycled, it might become possible for a user/
organization to unintentionally and inadvertently gain access to an information resource
that is not theirs. Allocation and deallocation of VMs, IP addresses, machine IDs, infor-
mation resources, and other such elements is the essential concern here.
Cloud service providers usually ensure this by allocating and deallocating IDs, IP
addresses, and other such elements using algorithms that make sure that such elements
are not repeated for a reasonably long time or until the previous NAT entry has been dis-
carded. The system makes sure that even if an IP address does get reallocated to some other
user, the NAT table does not have any information regarding the relationship between the
IP and its previous host. This enables rewiring the IP-host relationship from scratch.
Another cause of concern regarding multitenancy could be the data leaks, such as stack
overflows, that could possibly result from processing of data. There are multiple techniques
available to isolate a user's data from data belonging to other users. One of the solutions
is to assign a unique ID (based on the object being accessed) to each individual request,
thus identifying each data item separately. Isolation can be mutually reinforced at multiple
levels, such as at the physical disk level, the VM, disk partitioning mechanisms, and file
system permissions.
Cloud storage often uses centralized facilities, which can be a potential target for crimi-
nals or hackers. The threat can be mitigated by applying the appropriate security controls
(access controls, firewalls, and anomaly detection proxies). We will explore more on this in
the section “Cloud Security and Privacy” later in this chapter.
Complex hardware and software implementations form the storage systems. The poten-
tial is always present for catastrophic failure that might either destroy the data or even
worse expose the data from one customer to another, which is a breach of security.
To mitigate the consequences of such a failure, cloud service providers continuously back
up data for disaster recovery or retention purposes. Usually these backup data stores are
kept at an off-site and offline facility. The idea is that data should not just exist in one place
where it becomes vulnerable to operational failures that make it inconsistent and faulty.
Although providers act within the bounds of their contract and preserve the confidenti-
ality of these data copies, they are subject to error. They are also subject to jurisdictional
laws, which vary greatly from country to country and region to region. It is advisable to be
one step ahead and be cautious and vigilant about your data. We will discuss jurisdictional
laws later in this chapter.