The recent maturity of technologies such as high-powered multi-core CPUs, virtualization, and fast, ubiquitous Internet access has accelerated the adoption of cloud computing services. In fact, according to recent research, 50% of software applications being developed today are cloud-based. This continued shift to the cloud is impacting the network infrastructure of organizations as they rely more heavily on cloud computing systems for critical applications. Specifically, the characteristics of network traffic changes when it’s transitioned to cloud consumption models. And how those changes impact the quality of user experience (or Quality of Service) should be a priority for network teams. In this article, I’d like to quickly look at the different types of cloud services being adopted and provide some Quality of Service (QoS) best practices that NetOps teams should consider.
The industry focuses on three primary cloud services models: Software as a Service (SaaS), Platform as a Service (PaaS), and Infrastructure as a Service (IaaS). SaaS eliminates the need for end-users to purchase and manage computing infrastructure and software. PaaS eliminates the need to purchase and manage computing infrastructure and licenses. It also manages extra platform services like database and billing. In addition, PaaS can lock end-users into a specific platform environment. And IaaS eliminates the need to purchase and manage computing infrastructure and provides the flexibility to move to other platforms.
Most cloud services are delivered via the Internet using standard Internet client applications. Companies find this appealing since Internet access gives users the flexibility to use these services from virtually anywhere. The downside is that without adjusting QoS policies, standard Internet protocols such as HTTP and HTTPS are treated the same as casual Web browsing. Whether or not the anticipated load on the network will have an impact on the end-user is best determined in the context of the current network design and the services running across it. But any standard QoS policies that NetOps has in place will treat these applications as casual web traffic and assign them to a Best Effort Class. To ensure you have the proper QoS policies in place, here are seven key steps to consider:
1) Communicate with other IT teams: In larger companies, there are likely multiple teams involved with rolling out cloud services. For example, if the NetOps team is putting unified communications in the cloud, they might need to work with a different group than if implementing Google Apps for Business. It’s best that the team interject itself into conversations early regarding rollout rather than trying to fix things afterward. It’s helpful to ask questions about how best to identify traffic for prioritization and if there are any known concerns around the application’s fragility to loss, delay, or jitter.
2) Understand the traffic: NetOps teams should develop a good understanding of their current conditions by baselining network traffic. This is because cloud services will cause a shift in traffic and usage patterns by adding traffic to and from the cloud service provider. For example, Office and backup traffic that used to strictly be internal may now significantly increase the bandwidth demands on the network’s perimeter requiring circuit upgrades. Teams can baseline network traffic utilization by assessing specific applications consumption using a Netflow reporting solution.
3) Develop the initial design: It’s important to design how cloud services will fit into the current architecture and the corresponding QoS considerations (and policies). Like traditional WAN environments, cloud service traffic also fits best in QoS models with four or more classes. Minimally, real-time voice/video, transactional data, bulk data, and best-effort data traffic should be separated into unique classes for bandwidth management. In some networks, more might be needed for further segmentation. If baselining indicates that more bandwidth is needed or a change to the network architecture model, incorporate this into the plan.
4) Review the plan: QoS policies should not be done in a vacuum. Present the plan to other IT teams and key stakeholders within the organization. They will want to know about changes and how they might impact current services. If the existing network architecture needs updating (upgraded WAN services, new routers, or even a different network design), that will also need to be reviewed.
5) Design the details: Most cloud service traffic will use existing Internet protocols. Be sure to know the IP addresses of the services you will be connecting to in order to distinguish them from casual Internet traffic. Utilizing a NetFlow reporting solution can significantly help with this process. Some NetFlow solutions can also help look beyond basic IP address lists for traffic identification by utilizing application recognition features such as Cisco’s NBAR.
6) Implement and test: Test the new QoS policies in a lab or as a pilot deployment. Typically, the most difficult part of QoS projects is ensuring the traffic is being identified correctly and actually using the right queue. Utilizing NetFlow can be invaluable for ensuring this identification is happening properly. Sizing and buffer tuning are often easily achieved once identification has been successfully validated. This validates that everything is running correctly and gives the team an opportunity to fine-tune bandwidth allocations or reassign protocols to different classes if necessary.
7) Production and roll out: Implement the policies in a production environment and monitor and fine-tune as you go. To track key performance indicators of success or a problem, use NetFlow reporting and SNMP monitoring to ensure traffic is being identified correctly and high priority queues are not dropping traffic. This is also where a good network performance monitoring solution (NPM) can provide visibility and help fine-tune policies.
A final word
When deploying QoS for cloud services, or even investigating and troubleshooting performance issues with cloud services, having the right tools can help accelerate the implementation and troubleshooting of cloud services and provide a better end-user experience. Office productivity, offsite data storage and analysis, and application development and hosting are a few of the many operational efficiencies cloud services can provide. By planning for the changes in network traffic patterns and properly implementing QoS policies for these new cloud services, NetOps teams can avoid future headaches for both end-users and IT teams.
Jubil Mathew is Technical Engineer at LiveAction.