Blog

Monitoring and Operating a Private Cloud

July 24, 2018 | Ken Leoni

Many enterprises are under pressure to adopt cloud as an essential element of their IT strategy. An incremental approach that begins with a properly implemented and monitored private cloud provides a good foundation from which to build on. This is especially important to enterprises that can’t yet fully commit to the cloud due to technical constraints or a lack of time and resources.

A measured approach to cloud adoption should be a guiding principle as mistakes can be difficult to reverse and costly. While public cloud providers often tout all kinds of savings, a careful analysis needs to be performed in order to determine where and how those savings will be realized. 

Private, Public, and Hybrid cloud deployments each offer unique value propositions. Engaging in a proper capacity planning and cost analysis exercise is the best way to determine which cloud computing model will work best for your organization.

The costs associated with miscalculating a cloud deployment 

  1. Capacity

    • Private cloud – can be saddled with unnecessary fixed costs for hardware and software licensing.

    • Public cloud – the enterprise could be left paying for cloud resources that they can’t or won’t use. 

  1. Management and Monitoring

    • Private cloud – while virtual infrastructures are readily managed and monitored, movement to cloud-native applications is expensive as it requires a wholesale replacement/rewrite of the existing applications. In the end, a proper transition to cloud-native applications can be disruptive, time consuming, and expensive. 

    • Public cloud – operating cloud-native applications requires a fundamentally different set of skills to develop, deploy, manage, and monitor – skills which are in high demand and expensive even when used efficiently.

 

Where are you in your cloud journey?
Where are you in your cloud journey?

Private Cloud Stage 1


Private cloud - the enterprise is operating a private virtualized IT infrastructure (on-premises or collocated).

In this “infrastructure management” model the environment is relatively static with additional virtual machines provisioned by internal IT resources on an adhoc basis.

In many respects this is the re-branding of the traditional on premises data center with the cloud moniker.

Private Cloud Stage 2


Private cloud - the enterprise has fundamentally shifted from infrastructure management to a service delivery model ( i.e. 
IaaS, PaaS, or SaaS).

Platforms such as 
OpenStack or Azure Stack, provide the basis from which to leverage cloud-native applications.

Applications are considered to be “cloud-native” when they are 
containerized, managed dynamically, and operate as microservices.

Hybrid Cloud


Private and Public cloud operate together seamlessly, typically with cloud-native applications. The enterprise has extended the use of its cloud platforms to allow workloads to move between private and public cloud, giving them the ability to readily adapt to changes in cost structure and workload.

 Public Cloud


Critical applications are hosted by public cloud service providers with technologies such as AWS, Azure, or GCP.

The applications themselves could come from a virtualized infrastructure that has been migrated to the public cloud; the more likely scenario, they could be cloud-native applications that have been developed internally or provided by third parties that operate as Software as a Service.

Enterprises will often adopt multiple cloud platforms which introduces a whole new level of complexity, especially when it comes to operations and monitoring.

 

Monitoring and Operating Private Cloud

Private Cloud Deployments whether they are operating as infrastructure management or as a services delivery (IaaS, PaaS, or SaaS) all share a common set of characteristics that require close attention.

 

Physical Infrastructure

IT usually determines what type of servers, storage, and network infrastructure are physically installed as part of private cloud deployments (there are exceptions i.e. Azure Stack).

CPU Summary ReportThere is a lot to be said for the flexibility that private cloud deployments provide in terms of allowing IT to tailor the physical infrastructure to best suit the needs off their enterprise. IT also benefits from better control over the budget because the hardware life-cycle can be easily regulated as to what hardware gets upgraded and when.

Unlike public cloud deployments where computing, storage, and network resources are typically shared - the enterprise has exclusive access to its private cloud resources, which means IT is held accountable.

 

 IT is held accountable for the physical infrastructure


Hardware Health
– provisioning the hardware in a private cloud means that IT must monitor the physical health of the IT infrastructure. It is critical to be alerted to any hardware failures, especially ones that are transient (i.e. fan problems, power supply issues, disk array errors, etc..) before there is any downtime.


Server Performance – host resource utilization is especially important. Is the workload balanced? IT also must be able to readily identify future capacity issues (deficit and surplus) before users are affected.


Fail-over and Recovery - all cloud deployments maintain a certain level of resiliency. When a fail-over occurs, IT needs to be notified. Even though a successful fail-over should result in little or no disruption, IT still needs to restore the physical infrastructure back to a normal baseline.

 



Network Infrastructure

A private on-premises cloud offering provides a distinct advantage especially when hosted applications require low network latency or can’t tolerate internet performance issues. The monitoring of the network becomes especially important when we consider that applications usually span across multiple hosts and that a private cloud physical infrastructure (i.e. SANs ) are heavily dependent on network performance.

 

IT controls all things network


Network Bandwidth
– inbound and outbound traffic analysis per interface. Proactively identify patterns that could be a precursor to performance slowdowns.


Quality of the Bandwidth
– analyze for packet loss and re-transmission. Lost packets mean TCP re-transmissions, which results in more time required to get the data over the wire, which causes slow application performance.

 



Applications Performance

 

SLAUltimately it is about end user experience. Are critical applications available and responsive? Having a Service Level Agreement in place as part of a cloud monitoring strategy can go a long way towards identifying what, if any application performance problems happened, where, and when.


 Service Level Agreements measure application performance

What is the availability of the application? Availability is typically a heartbeat test: checking to see if the application and its components are alive. Testing can take the form of pings, port checks, launching URLs, or any other test that establishes the service is available. 

 

How is application performing? Verify that the application is functioning and responsive via synthetic web transactions. Confirm that expected content is returned and that the response time is within acceptable limits.

 

What is root cause of performance degradation? By aggregating underlying IT infrastructure and application components of the application into an SLA, IT can readily determine the root causes of performance problems.



 

Summary:

 

A number of factors enter into the equation when implementing any private cloud monitoring strategy. Certainly, targeting physical, network, and application performance are critical to a successful deployment. However, perhaps the biggest wildcard that can ultimately govern success or failure is the time and resources that IT has to properly plan for and execute a private cloud strategy.

While a private cloud based on an infrastructure management model offers a fairly predictable cost model; a private cloud that offers cloud-native applications can be more difficult to cost out and more expensive to operate.

 

 Infrastructure Management vs Cloud-Native Private Clouds


Infrastructure Management - the private cloud infrastructure tends to be static in nature. The abundance of virtualization, network, and application expertise means finding the right people to implement is not particularly challenging.


Cloud-Native - Enterprises that adopt a private cloud delivery model of IaaS, PaaS, or SaaS must also adopt a new DevOps methodology. Not only is application development quite different, but the skill-sets required of IT to properly deploy and administer are unlike those of infrastructure management.

For example, OpenStack although quite capable, comes with a significant learning curve – therefore talent comes at a premium.

 



Ultimately an enterprises successful implementation of a private cloud management and monitoring strategy may be governed as much by finding the right personnel or MSPs as the choice of the underlying technology itself!

 

Want to learn more?

Download our Virtualization or Cloud IaaS Whitepaper -  While Cloud IaaS and virtualization technologies have common underpinnings, there are significant differences between them that can make one technology a better fit over the other.

Download the whitepaper:  Virtualization or Cloud IaaS?

We value your privacy and will not display or share your email address

Sign Up for the Blog

Heroix will never sell or redistribute your email address.