Cloud Smart’s success depends on smart network monitoring
- By Mav Turner
- Apr 11, 2019
The Office of Management and Budget’s Cloud Smart proposal signals both the end of an era and the beginning of new opportunities. The focus has shifted from ramping up cloud technologies to maximizing cloud deployments to achieve the desired mission outcomes.
Certainly, agencies are investing heavily in these deployments. Bloomberg Government estimates that federal cloud spending will reach $6.5 billion in fiscal year 2018, a 32% increase over last year. However, all that investment and potential could be for naught if agencies don’t take a few necessary steps toward monitoring and troubleshooting distributed cloud networks.
1. Match the monitoring to the cloud. Different agencies use a variety of cloud deployments: on-premises, off-premises, and hybrid. Monitoring strategies should match the type of infrastructure in place. A hybrid IT infrastructure, for example, will require monitoring that allows administrators to visualize applications and data housed both in the cloud and on-premises. It requires a much different and more robust approach than a network-based or a private on-premises cloud.
2. Gain visibility into the entire network. It can be difficult for administrators to accurately visualize what’s happening within complex cloud-based networks. For instance, in a hybrid IT environment data passes between the cloud provider and on-premises operations. It can be tough to see what’s happening when that data is being managed outside of the organization.
Administrators must be able to visualize the entire network so they can accurately pinpoint the root cause of problems. Are they occurring within the network or the system? Are they related to a particular on-premises connection, or are they the result of a service provider issue? These questions need to be answered to maintain a functioning and well-optimized distributed network.
3. Reduce mean time to resolution. MTTR describes how long it takes an administrator to fix a problem from the time the issue is identified to when it’s resolved. A lower MTTR means more uptime, efficiency and productivity.
Data visualization and aggregation can be useful in minimizing MTTR when a problem arises. Administrators with rapid access to network, system and user information can easily and quickly correlate the relevant data , overlay it and accurately determine the issue. That’s much better spending the time to go to three different teams to solicit the same information, which may or may not be readily available.
4. Monitor usage and automate resource lifecycle to control costs. Agencies should carefully monitor their cloud consumption to avoid unnecessary charges that their providers may impose. Those charges could be related to incorrect billing or to the administrator not understanding what services are free. They should also be aware of costs and monitor usage of services like application programming interface access. Often, this is free -- up to a point. Being aware of the cost model will help admins guide deployment decisions. For example, if the cost of API access is a concern, administrators may also consider using agent-based monitoring, which can deliver pertinent information without having to resort to costly API calls.
The other key to keeping costs down in a government cloud environment is ensuring a tight resource lifecycle for cloud assets. Often, this will require automation and processes to prevent resources from existing beyond where they are needed. IT shops that are manually provisioning resources, must also manually de-provision those resources to stop the charges. Just because admins think they're no longer using a service doesn’t mean it doesn’t still exist, running up charges and posing a security risk. Tight control of cloud assets and automated lifecycle policies will help keep costs down and minimize an agency's attack surface.
5. Ensure an optimal end-user experience. Proactively monitoring end-user experiences can provide real value and help ensure that the network is performing as expected. Periodically testing and simulating the end-user experience allows administrators to look for trends that could signal the cause of network problems (periods of excessive bandwidth usage, for example). Knowing these trends ahead of time can help administrators better plan for potential network slowdowns.
6. Scale monitoring appropriately. Although many government projects are limited in scope, agencies may still find they need to scale their cloud services up or down at given points based on user demand. Monitoring must be commensurate with the scalability of the cloud deployment to ensure that administrators always have a clear picture of what’s going on within their networks. Agencies should build network monitoring capabilities for today while always keeping an eye on the future.
Successfully realizing Cloud Smart’s vision of a more modern IT infrastructure based on distributed cloud networks will require more than just choosing the right cloud providers or type of cloud deployments. Agencies must complement their investments with solutions and strategies that allow them to make the most of those investments. Adopting a comprehensive monitoring approach that encompasses the entire cloud infrastructure is the smart move.
Mav Turner is senior director, product management, for SolarWinds.