Scalable and virtualised cloud computing proves a challenge

Ian Cummins discusses the challenges faced when moving to the cloud

Cummins into the cloud

In tight times, a move to cloud computing might seem the obvious answer, especially with the introduction of flat-rate billing and flexible tariffs removing the consumption-per-use anomaly.

However, such an apparent gift horse does not always deliver. A scalable and often virtualised computing environment brings trials and cultural changes.

Unless the setup, operational challenges and minutiae are understood and addressed, a cloud resource will not be effective. Employees will face poor quality of service and lack of network capacity and application choice. Partner and customer access could also be lost. Security will become a problem, especially if the cloud provider is lax about data storage.

Ensure your existing network infrastructure can handle the bandwidth demands of cloud-based applications. Set up application use policies to prioritise traffic and ensure application reliability and network performance.

Prepare the IT team for changes in network management focus. Prepare the wider workforce for changes in business processes and culture – all employees need to understand the benefits of the move from a personal, as well as business, perspective.

Have contingency plans for scheduled and unanticipated internet outages. Do not be held hostage by your ISP. Have strict, accountable SLAs.

Ensure your cloud provider offers the freedom to install new applications without complex network administration and contract restrictions. Where a complete virtual environment is set up, be sure your service provider has adequate storage facilities.

You need internal backup provision and to use the most appropriate cloud provider for your operational needs.

The key benefits at this point are the ability to scale services and applications quickly to meet changing business and user demand, while avoiding the capital cost of expensive network assets for intensive but infrequent computing tasks.

However, this is where the crunch comes. By providing increased IT flexibility, cloud computing shifts IT management priorities away from the network core to the point of WAN/internet connection. This extends the network beyond the safe confines of the corporate firewall.

Plan, deploy and manage
Determine existing bandwidth demands per user, per department and the organisation as a whole. With the service provider’s help, calculate the average bandwidth demand per user for each new service you plan to deploy.

The IT team can scale the internet connection and prioritise and shape traffic to meet the bandwidth demands of cloud applications.

Maximise the cloud computing advantage by pushing the application, data storage and processing burdens on to a virtual network.

Shift management priorities from internal to external data concerns. This will significantly affect the decisions made – from network monitoring and WAN performance tracking through to the personnel and resources to be devoted to managing WAN-related issues.

Use cloud and SaaS applications to move bandwidth demands away from the core network where processing power is focused.

Data prioritisation becomes vital when reliance is placed on a single pipe responsible for handling the vast majority, if not all, of an organisation’s applications and processing.

If a single IP consumes 30 per cent of the available bandwidth, it becomes unworkable. Therefore, applications must be prioritised and traffic throttled to ensure capacity is best allocated.

Assess the reliability of the existing ISP, its SLAs and service assurance. Be aware of scheduled maintenance and consider the feasibility and practicality of having multiple providers to dispense backup should downtime or performance become an issue.

SLAs must take on greater importance in a cloud environment. Unlike a standard network environment where traffic can be monitored from client to server, immediate problem identification and resolution is far more complex in the cloud because most data is controlled by the service provider. This makes it extremely difficult to monitor, optimise and troubleshoot connections.

These SLAs should specify expected internet service levels and performance obligations the service providers must meet, and define unacceptable levels of dropped frames and other performance metrics.

It is not in the provider’s interest to tell a customer when service drops – or fails completely. Therefore, it is vital to have a network analyser in place to provide an independent overview of WAN link connections.

The most effective way to do this is to use a network analyser with a WAN probe to verify service levels. This will provide management reports that show whether or not the provider is meeting its obligations.

An SLA is not enough to guarantee your level of service. Security is a major concern. Ask:

• Who within the cloud’s employee base has access to data?
• How is a customer’s data segregated?
• Is it encrypted?
• What happens in the event of a disaster or security breach – who is responsible for data recovery and restoration, and how long will this take?
• What happens to the data if the cloud provider goes out of business, and what provision, if any, is put in place to provide alternative application resources?

Cloud computing is no longer a buzzword – it is a realistic way for organisations to have network flexibility, scalability and computing power for minimum investment.

Like most technologies, these services are not without risk and require proper preparation and management to succeed. If a network is adequately prepared and the IT team works with the management to support change, they will realise the true promise of cloud computing.

Ian Cummins is EMEA vice president of Network Instruments