Before you deploy cloud, you need to decide which cloud management tools you will use. There are multiple sources. In virtualization suites, some tools are included natively, for example. There are also third-party tools that promise single-pane-of-glass management across numerous distributed data centers. As we’ll examine later, there are pros and cons to using each kind. Your cloud management tool choices should be informed by the needs of your specific workloads.
As with any technology, the ability to monitor cloud infrastructure in conjunction with other interdependent components will dictate just how robust an environment should be. Private, public, or hybrid cloud may each require its own set of tools.
But there will be common considerations among all major cloud management tool sets. With major infrastructure components, administrators must have clear visibility into their environment. Good tools and monitoring software should include the following features:
- User count. At any time, an administrator must know how many users are accessing the cloud environment, which server these users are on, and which workloads they are accessing. This type of granular control allows IT administrators to properly balance and manage the server-to-user ratio. The only effective way to load-balance cloud servers is to know who is accessing them and in what number.
- Resource management. Deep resource visibility comes on multiple levels. As discussed, it’s important to see how physical resources within the cloud are used. This also means viewing graphs, gathering statistical information, and planning for the future. Visibility and management revolve around an administrator’s ability to see which resources are available, and where they are allocated. Improper allocation can quickly become too expensive.
- Alerts and alarms. A healthy environment with good cloud visibility includes alerts and alarms to proactively catch issues. By catching problems before they become outages, an organization can maintain higher levels of uptime. Being able to set up alerts so that the proper administrator is notified depending on the issue is important. If a storage alert goes out to a server admin, the response may not be as fast as it would be had the alert gone out to a storage administrator.
- Failover capabilities. With good visibility comes the ability to fail over cloud servers without creating user downtime. If an error or issue is caught, administrators can fail users over to a host capable of handling that volume. In many environments, this task can be automated. If a physical host goes down, the VMs residing on the host will be safely migrated and balanced among other available servers, with alerts sent out to appropriate parties.
- Roles and privileges. Good visibility also means having roles and privileges built into the environment. This means that the storage team has access only to cloud-based storage components, and the virtualization team can have access to VM management. This isolation of roles creates effective audit trails. It also greatly reduces the risk that a team member will make the wrong changes to the system.
- SLA considerations. When working with a third-party provider, visibility into service-level agreements is also critical. This means monitoring uptime and environment usage. Depending on the type of SLA, different metrics are important to the administrator. This might mean monitoring the number of VMs working or adjusting downtime requirements.
- Testing and maintenance. Just as with any other infrastructure, cloud environments require maintenance and testing. Tools that help administrators with server patching, updates, and other general maintenance tasks are valuable. Planning for testing of bandwidth or failover capabilities must also be in place.
Most of all, it’s important to make sure your cloud management tool set is directly aligned with your data center strategy and your business goals. Remember, your underlying infrastructure is the main driver for your entire business. Without the right management tools your go-to-market strategy could suffer.