Cloud computing is promising much – but is failing in many areas as users get to grips with some of its more complex areas. An example here is when organisations start to look at how best to use multiple cloud platforms across a private and public environment – what is known as a ‘hybrid cloud’.
In essence, such a cloud sounds easy: an organisation maintains certain workloads on its own equipment, using public cloud where and when suitable to meet the needs of a process.
However, the devil is in the detail. The first problem starts where the choice of technical cloud platform has to be made. Using different cloud technologies can make workload and data mobility difficult. For example, using an OpenStack private cloud and a Microsoft Azure public cloud means that compromises must be made in certain areas.
However, Microsoft has now addressed this with the launching of Azure Stack – a highly engineered hardware/software system that can be made available to organisations to create a private cloud that is highly compatible with the Azure Public Cloud.
Great – a major step forward. However, the next problem comes with performance. Having an Azure Stack instance in an organisation’s own datacentre and then trying to connect through to the Azure Public Cloud means that data (and, in many cases, application workloads) has to traverse across a wide area network (WAN) connection.
If this is a public internet connection, then performance will generally be bounded by a ‘best efforts’ constraint. If it is a dedicated link, then costs may be a problem.
Again, Microsoft has tried to address this, introducing its own dedicated connections into the Azure Public Cloud. These connections are offered under the ExpressRoute banner, and provide ultra-low latency, high bandwidth paths into Microsoft’s cloud facilities.
Wonderful – except that these connections do not terminate in a general customer’s premises. To achieve suitable scalability within cost constraints, Microsoft has struck deals with other facility providers to offer points of presence (PoPs) for ExpressRoute terminations. End customers then need to strike a deal with these PoP providers, who will then provide dedicated connections using quality of service technologies to connect into the end customer’s facility – so there is still the issue of this last link that must be monitored and managed outside of the Microsoft environment.
Such complexity is hard to avoid, but can lead to problems for those aiming for a seamless logical hybrid cloud platform. Luckily, there is one way around all of this: the use of a colocation provider who is also an Express Route termination point.
Here, the end customer takes space within the colocation facility and places their Azure Stack equipment within it. Using intra-facility connectivity speeds, they then connect through to the ExpressRoute environment, giving an Azure Stack/Azure Public Cloud experience that is essentially one consistently performing platform.
By choosing the colocation provider carefully, it can be possible to further open up the options around connectivity. Good colocation providers will have both dedicated, high performance networks underpinning connectivity between their own and other facilities, but will also offer direct connectivity into the main public clouds.
By ensuring that a colocation provider has the right portfolio of services, end user organisations can ensure that they maintain flexibility over how they continue their journey into a fully-functional hybrid cloud, and can make the most of the opportunities that such a platform can provide.
Quocirca has written a short report, commissioned by NGD, on how collocated Azure Stack using ExpressConnect can provide the best option for organisations wanting to utilise a hybrid cloud. It can be downloaded for free here.