Avatar

This blog post is a collaboration between Vishwanath Jakka, Engineering Product Manager, Cisco Cloud & Compute & Co-authored by Francoise Rees, Product Marketing Manager, Cisco Cloud & Compute.


Cisco Workload Optimization Manager (CWOM) recently introduced support for Google Cloud Platform (GCP), providing visibility, insights, and actions for the top three public cloud providers. It’s no secret that many organizations use multiple public cloud providers – either by design, to take advantage of a provider’s specific strength or toolkit, or due to mergers and acquisitions, where different business units use different cloud providers.

According to the Flexera 2022 State of the Cloud Report, 89% of organizations have a multi-cloud strategy, with 80% having a hybrid cloud strategy. Using multiple clouds just compounds the challenge of controlling cloud spend.

HashiCorp, in its 2021 state of cloud survey, also found that 39% of respondents said their organization overspent on their cloud budgets.

CWOM to the rescue

Cisco Workload Optimization Manager is a decision-automation engine that provides workloads with the exact resources they need at the right time in accordance with business policies, freeing IT teams to focus on innovation rather than the maintenance of their IT environment.

CWOM helps organizations assure application performance and reduce cloud cost by continuously resourcing applications to perform better. By using CWOM, customers let software make the decisions through AI and automation to continuously assure applications have the resources they need to perform at the lowest cost. CWOM scales the efforts of your IT and application teams, helping them drive innovative applications and differentiate against the competition.

And now with version v3.3.0 of CWOM, IT teams will have insights into GCP the same way they already have into AWS and Azure. Starting with this version, CWOM will assist customers in extending a performance-centric view to workloads on the Google Cloud Platform (GCP), like we do for AWS and Azure. CWOM gathers specific data points from GCP and provides actionable and automatable recommendations to maintain and increase performance, while reducing costs.

The first step is to register your GCP accounts during setup. Once a Google Cloud Platform Service account has been added, the Workload Optimizer Manager will automatically discover the infrastructure information.

Adding a Google Cloud Platform Billing account will enable CWOM to use cost data analysis capabilities to provide accurate recommendations for your workloads.

Data Gathering

Once target registration is complete, CWOM collects a set of data points needed for the analysis.

  • Projects – CWOM discovers GCP projects that define compute, storage, and networking resources for the workloads. With organization-level permissions CWOM will discover full resource hierarchy (organization, folders, and projects) allowing for an in-depth and holistic analysis of your environment.
  • VM Resources – CWOM will automatically identify and gather data about the resources utilized by each virtual machine: Virtual memory (vMem), Virtual CPU (vCPU), attached storage, IOPS and throughput.
  • Billing Data – By adding the billing account CWOM will have access to GCP bill and service costs over time. Negotiated discounts and GCP’s committed use discounts are also collected.

Recommendations and Actions

Once the data points are collected CWOM starts the analysis to provide recommendations and actions. The two types of actions introduced are Scale VM and Reconfigure VM. We will expand the types of actions provided in future releases.

  • Scale VM Actions – CWOM takes the VM resource utilization data, workload costs, and scaling constraints to recommend actions to optimize performance for the workloads and business applications.
  • Reconfigure VM Actions – GCP supports a specific set of machine types for each zone in a region. If you have defined policies that restricts VMs to certain machine types and the zones they are currently on do not support those machine types, CWOM will recommend a reconfigure action to notify of the non-compliant VM. For example, if a policy restricts a VM to M1 family machine types but the zone does not support machine types for the M1 family, CWOM will recommend VM reconfiguration.
Cisco Workload Optimization Manager demo screenshot of GCP Workloads
Figure 1: GCP Workloads visibility: List of VMs sorted by status

 

Cisco Workload Optimization Manager demo screenshot of changing instance types
Figure 2: Scale VM action: Change instance type to relieve read throughput congestion 

 

Expanded Stack Coverage

Now Workload Optimizer Manager, with support of GCP, can provide visibility and insight for both the virtualization and containers layers of the stack. CWOM has supported Google Kubernetes Engine (GKE) for a while. Now with GCP info, Workload Optimizer Manager can stitch GKE clusters to GCP resources for holistic visibility and actions.

  • Cost Visibility – Customers now have a better understanding of the costs associated with their GKE clusters. If the clusters are multi-tenant, this cost visibility can be used for show back. Clusters can scale horizontally to support more and more applications. When that happens, you see the associated GCP costs and can understand the investment required overall.
  • Unused node management – This will also enable turning off unused nodes confidently. CWOM removes the guess work from node management by making sure that the active workloads can run effectively on the remaining nodes. By stitching GKE to GCP resources, cost savings are realized by using only the resources needed to run your workloads.

In summary, with the newly released support for Google Cloud Platform, Cisco Workload Optimization Manager now lets you manage and optimize workload resources across the top three public providers, to help you reap the benefits of a well-tuned cloud environment.

 


Resources

Send an email to trycwom@cisco.com to get started on your CWOM free trial

Access our CWOM document hub



Authors

Vishwanath Jakka

Product Manager, Compute and Cloud

Compute and Cloud Group