Organizations making the move to cloud services are well aware of the many benefits like lower startup costs, lower total cost of ownership and on-demand scalability. But many companies fail to take a hard enough look at the corresponding challenges associated with application availability and performance.
For many cloud services deployments, network performance will be the key to application performance. Why? Because every business service accessed over cloud infrastructure is by definition a remote, network-dependent service. Every user becomes a remote user and every office, even the corporate headquarters, becomes a remote office.
Even services-based applications will stumble, freeze and eventually disconnect users when network bandwidth, jitter, latency or packet loss metrics drop below acceptable thresholds. Many of the applications being moved to cloud infrastructure were not originally designed for remote access, making them even more susceptible to changes in network performance.
In short: a successful cloud services deployment depends on your ability to guarantee application performance from the user’s perspective. That means you need to manage network performance. What are the key steps for managing network performance for cloud service deployments?
1. Network assessment
The first step in ensuring a successful cloud services deployment is to perform a comprehensive baseline assessment of network performance. When rolling out cloud services – or other IP-based services like VoIP, video conferencing, virtual desktop infrastructure (VDI) or IP storage – many organizations underestimate the operational and business risk associated with unplanned network impacts.
A network assessment is the only way to accurately know the scope of the project and its costs. It’s also the only way to know if your network is up to the task of carrying the extra traffic.
2. Continuous monitoring
Users need the performance of cloud-based services to be at least as good as what their current infrastructure delivers. To guarantee acceptable performance you need to be aware in real-time of the ever-changing status of the networks connecting remote users to applications. Latency, jitter and other key network performance metrics can fluctuate continuously in response to changes in traffic and other factors.
Continuous monitoring capability is also required to give you real-time visibility into service levels. This is the basis for troubleshooting network problems, as well as for ensuring that cloud and other third-party providers are meeting their contractual commitments.
Further, the physical distance that data must take between the cloud and the user has a huge impact on performance, especially for TCP based applications. This parameter will vary among different public cloud providers, making monitoring performance over time an important step in evaluating cloud hosting options.
3. Proactive troubleshooting
Today’s cloud-based services are business-critical and ensuring their availability is vital. Traditional break-fix approaches to network management are not adequate to meeting many of today’s SLAs. Network engineers therefore need a way to pinpoint the exact cause and location of performance degradation — even within the virtual network and servers making up the cloud.
This makes it possible proactively address network issues before they impact users. Likewise, if cloud services crash and a plethora of service providers start finger-pointing, you need to know, quickly and decisively, who is responsible for what.