Full-Stack Infrastructure Observability with DX Unified Infrastructure Management

The IT infrastructure landscape has seen tremendous changes over last few years due to evolving technologies, newer business models, and ever-changing market demands. Business, market, and consumer demands are pushing such IT advancements as cloud, mobility, and IoT. Specialized IT infrastructure monitoring tools provided by original equipment manufacturers, the adoption of automation, and advanced AIOps tools that break down silos in the observability space are enabling these advancements in modern data centers. We have entered into an era in which tools and automation matter far more than they used to compared to the traditional service-oriented IT environment.

Monitoring that offers visibility and deep insights into IT infrastructure is imperative for any organization. Contemporary enterprises are exploring infrastructure monitoring tools that predict anticipated issues and prescribe the solution to mitigate those issues.

IT infrastructure monitoring has shifted from traditional reactive support to an era of predictive monitoring and support, leveraging tools that predict what’s going to happen in the future. Predictive monitoring is gradually moving towards prescriptive monitoring, a state in which advanced analytics will provide insights into future possibilities and challenges within the IT infrastructure.

Figure 1 : Modern hybrid data center

The Modern Hybrid Data Center

With ongoing digital transformation projects, IT operations (ITOps) teams must now manage the complex infrastructure stack underpinning business services in the organization, as seen in figure 1. Today’s data center is a mix of traditional infrastructure components and modern cloud assets working together to deliver the digital experience.

Application Infrastructure Layer

SaaS applications like Microsoft Office 365 or web applications hosted on web servers like Apache and IIS comprise this layer of the modern data center. Virtual desktops like Citrix also need to be managed. Synthetic end-to-end (E2E) application monitoring can be used to track the end user experience delivered by application services. Some enterprises have built their data pipelines on the Hadoop 2.0 ecosystem using Cloudera or Hortonworks distribution, and providing data ingestion, processing, and presentation services based on RDBMS or no-SQL data stores.

Infrastructure Layer

The infrastructure layer in figure 1 covers multiple technologies suited for different tech stack elements, which support the application infrastructure by providing computing, storage, and memory. This layer delivers computing services through:

• Physical or virtual devices, which may be running on Windows or UNIX.
• Containers like Docker, Kubernetes, or OpenShift for private clouds.
• Cloud service providers with PaaS or IaaS offerings.

PaaS or IaaS cloud computing reduces capital expenditures and provides highly elastic resources for hosting new age applications through subscription-based licensing models. Lately, large enterprises are receiving large bills that impact operating expense budgets. To avoid billing shocks but continue to provide the required resources, large enterprises are shifting to hyperconverged infrastructure software from vendors like Nutanix and VMware.

Another key component of the infrastructure layer is storage, which can be primary or secondary depending on the end customer use case. NetApp, EMC Unity, Cohesity, and Isilon are popular storage systems that provide low-cost persistence.

Network Layer

The network enables communication across application and infrastructure components. Every user can verify the connectivity and availability of network switches, routers, or any other network layer component over SNMP or REST endpoint. For cloud-based deployment, the network layer also has a virtual private cloud (VPC) for connectivity, which needs to be monitored in conjunction with other components in the stack.

Personas

The personas supporting the full stack are changing based on evolving roles and responsibilities. Server, storage, and infrastructure administrators are becoming Site Reliability Engineers (SREs) responsible for full stack availability and reliability. To manage the scale of inventory, operations teams are adding specific roles like maintenance engineers, tools administrators, and operation engineers. Because intelligent fault management has led to alarm noise reduction, bots and rules on IT service management (ITSM) tools like ServiceNow or CA Service Desk are replacing L1 support engineers. L2/L3 infrastructure specialists are gaining importance due to their ability to triage infrastructure observability issues.

Figure 2: Roles and personas in the modern data center

Key Requirements

With an expanding infrastructure stack and evolving key personas within data centers, there are several key requirements of a unified infrastructure monitoring solution. To provide effective and efficient monitoring in the modern organization’s hybrid estate, teams need a solution to:

• Provide full-stack infrastructure observability across applications, infrastructure, and network, including cloud infrastructure and networks.
• Monitor service levels across the private and public cloud. It must also be able to define service-level objectives (SLOs) based on quality of service (QoS) parameters. Similarly, it must also be able to define service level agreements (SLA) based on these SLOs.
• Check cost and resource utilization with out-of-the-box billing and health reports.
• Ease triage with technology-led dashboards that can be both out-of-the-box and customized based on triage needs.
• Offer quicker time to value with a template-based monitoring configuration service (MCS) for a dynamic yet standardized bulk monitoring configuration across technologies.
• Manage inventories and groups with CMDB integration and dynamic discovery.
• Deploy across various infrastructures, such as on-premises, SaaS, and MSP.
• Integrate with IT operations management (ITOM) tools for provisioning and orchestration through published APIs.
• Reduce alarm noise with fault management—along with incident management—through ITSM.
• Provide an open, multi-tenant architecture that can meet a global enterprise’s needs and also host centralized monitoring offerings from managed service providers.

Broadcom’s DX Unified Infrastructure Management (DX UIM) meets all of these requirements. DX UIM offers a “single pane of glass” for monitoring across the stack, providing a unified solution for hardware and software management. It serves as a complete solution for monitoring performance, managing alarms, and enabling service quality degradation management. DX UIM offers comprehensive hybrid IT management on par with on-premises capabilities, delivering consistent, top-to-bottom infrastructure observability.

For more information, please visit the DX UIM home page at the Broadcom Enterprise Software Academy.