virtualisation & cloud services
of cloud should provide the ability
to burst (ie. scale up and down) as
needed either automatically or through
a simple self-service mechanism.
If your provider can’t do this it’s not
really cloud. You should only pay for
what you use, and be able to add or
reduce capacity, and associated costs,
as needs change. If your organisation
needs 100 virtual servers for a month,
they could be provisioned and ready
for use within 48 hours. You simply
turn the capacity on and then turn it
off when you no longer need it, only
paying for the time used.
Second, many organisations think
that they have to buy a standard
package. However, the better
managed cloud services can be
customised to your organisation’s
characteristics. You can combine this
type of service with ‘vanilla’ services
to create your own hybrid cloud to suit
your needs.
Finally, there are now standards
in areas such as interoperability, web,
authentication etc., which means
that some of the historical integration
challenges can be avoided and these
will help to increase the spread of
hybrid cloud services.
The key challenges in
implementing hybrid cloud are around
integration and standards. At present
not many people are dynamically
moving workloads between
organisations. Once you have chosen
to provision services with different
providers you will need the ability to
integrate those services. Toolsets to
do this and remote management and
reporting capabilities are also evolving
within both the commercial and
open source worlds. When defining,
running or buying services you should
make sure that the interfaces used
are as standard as possible e.g. XML,
SOAP, REST, SAML etc.
Figure 2: Service integration and management (SIAM) for hybrid cloud
Managing and monitoring
multiple providers
Once you have a portfolio of cloud
services, you need to actively
manage that portfolio and monitor
performance against the agreed SLAs
to ensure you receive the contracted
service. You will want to know:
How well are my service providers
performing against contractually
agreed SLAs?
If they are not performing, where
is the problem? This is particularly
important where multiple providers
are responsible for elements of the
IT service
Is the aggregated service
delivering suitable performance to
our user community?
This is leading to a growth in
new services (Cloud Monitoring as
a Service, or CMaaS) to monitor the
performance of multiple suppliers, all
of whom will claim ‘it’s not their fault’
when a problem arises. An effective
monitoring service should pull
together service availability and other
performance information. It should
have the ability to carry out synthetic
transactions against defined services
and applications, show overall system
health, and monitor response times
and latency. It should also consolidate
events and other performance
statistics across the IT supply chain.
Management services are continuing to
evolve to meet the needs of hybrid cloud.
Ideally, an organisation would
carry out this monitoring from a
single pane of glass. Fordway has
developed a solution with its CMaaS
service, taking publicly available
information from cloud providers such
as Amazon Cloud Watch and using
the AWS integration tool plug-in
for the company’s toolset, which is
based on MS Systems Centre. The
same can be done for other public
cloud services such as Azure, Office
365, Service Now and Salesforce.
Network monitoring is also provided
25