Cloud Resilience and the Quest for Higher Availability
February 7, 2014
Virtualized networking environments have evolved to deliver cost effective and more consistent networking performance to end users. Not until recently however, the terms ‘high availability’, and ‘cost effectiveness’ were seldom mentioned in the same sentence. Skepticism surrounding cloud services used to revolve around the notion that cloud platforms were neither secure nor resilient enough to deliver superior application performance during times of peak traffic.
Thankfully, times have changed as data centre companies have devised a solution with AWS Direct Connect offering end users higher availability via private network connections. AWS Direct Connect has the ability to keep bandwidth costs down because it isolates public and private internet environments between specific locations.
Suddenly, private cloud platforms can deliver greater availability than the public internet can offer. What about the reassurance of consistent networking performance?
Cloud resilience of any type is dependent on the hardwired infrastructure from which it is based and what happens when data travels from point A to point B.
In this case, it's the Data Centre - where numerous ISPs (internet service providers) come together to exchange and route traffic based on the behavioral flow of data. 'Peering' as it is known, allows separate internet networks to interconnect and exchange traffic to and from a server on demand. 'Peering', respective to every ISP, also helps retain value amongst those who are utilizing the private network more often and without availability problems. Data Centre companies provide the secure hosted environment for cloud customers and offer them the following benefits:
Private connectivity to balance bandwidth workloads
Redundant power and disaster recovery
Scalable cost structure for greater flexibility
Compatibility and Access to AWS services
Availability Vs Uptime
When cloud customers say they are looking for reliability they really mean to say they are looking for high availability, but availability and uptime can sometimes mean two different things. When a cloud provider or data centre hosting company says they offer 99.9999% (or more) uptime, what they really mean is the time in which a cloud server is 'powered on' and available to system administrators. 'Uptime', in this case pertains only to power. In some cases, multitenant datacenter providers may use the term uptime and availability interchangeably to indicate guarantee of its data centre infrastructure and interconnection services. Like with Telx, it defines this SLA for interconnection is stated as "The service availability goal is 100% uptime."
'Availability' typically implies a broader connotation that includes both availability of the physical environment (i.e. server power and network within the data centre) along with the external network connecting users to the physical infrastructure. What end-users of an application or system care about is if the services are available or not, along with the response time or performance from an end-to-end perspective. So where does this leave us? We obviously know that cloud resilience is bolstered by peered connectivity, scalable bandwidth and a hosted environment resistant to outages, but what good does it do if a company is continually subjected to a public cloud network? They still suffer from availability issues.
With specific private cloud networks, especially amongst AWS Direct connect users, availability increases and stays consistent, even during peak hours of web traffic. This is because clients are hardwired from their point of presence straight to the data centre via a dedicated cross connect. When higher than average availability is desired, public internet connections simply do not deliver because once so many people bombard the network, congestion ensues and that causes latency.
Redundancy will always be of significance within any interconnected environment susceptible to power outages. The redundancy power components of a data centre ensure virtualized networks function during an outage allowing server resources to be available to the end user. Uptime is really a given in today's Tier III market, so availability is all the end user really cares about.
AWS Direct Connect is but one tool an organisation can use to attain higher availability. If by paying for a private direct connect, routed through a redundant infrastructure is what it takes to achieve cloud resiliency and higher availability, the investment may end up paying for itself.