Share your experience using CloudStack

We are selling a service based on CloudStack, utilizing it to sell resources, specifically virtual clouds, to our customers.

While it is accurate to say we are not a reseller since CloudStack is an open source product, we are indeed a user of it and are reselling services based on CloudStack.

Right now, we are using CloudStack with the KVM hypervisor, but we also have a couple of VMware clusters and are going to add a Proxmox cluster to our CloudStack as well, utilizing three different hypervisors based on the customer's requirements. We use different classes of hardware so we can offer tailored solutions, whether that means many cores with lower core speed or fewer cores but faster threads for database needs.

The flexibility in CloudStack is one of its best features, and while it's very flexible in how you can configure it, this flexibility can also lead to complex solutions, so you have to be very careful in how you implement your solution because you can do things in many different ways.

The advanced network management capabilities in CloudStack are fairly robust, allowing you to create virtual clouds, set up various types of VPNs, and perform many different networking functions, so it's pretty well-featured in networking.

The user management in CloudStack is very flexible, allowing you to use role-based access control and configure users, accounts, and domains in many different ways, building various structures for users and supporting different authentication methods such as Active Directory and LDAP.

The APIs for automation in CloudStack are excellent; everything is API-driven, and while the API is very rich and hard to learn, you can do everything you would want to do, including everything in the user interface and more. Furthermore, there is a very good CLI called CloudMonkey that utilizes the APIs for scripting processes, making automation very rich.

There is a need for improvement, particularly in the user interface for end users, which is fairly complex. Users require a good understanding of VMs and networking to utilize it effectively, so it might be beneficial to have a simpler interface for deploying a user and a VM while automatically setting up necessary configurations in the background. That said, you can create your own front-end using the APIs.

Our use of CloudStack has been ongoing for about two years.

The installation of CloudStack can be quite challenging because you need a good understanding of the product before starting. While setting up a very simple CloudStack with a few computers is manageable, attempting to configure extensive networking and various classes of computers makes the installation process very complex.

The challenge in installation comes from the complexity that arises due to CloudStack's flexibility; if you're not well-versed in what you're doing, it's easy to implement it incorrectly and end up with a fragile solution, similar to installing complex solutions like VMware, which can also become complicated if not managed properly.

From what we've seen so far, CloudStack appears to be fairly stable, showing no significant limitations.

CloudStack is also very scalable; you can build fairly large clusters without issues, making its scalability a strong point.

I have not worked directly with something similar to CloudStack for the same use case, although we did evaluate other solutions, including OpenStack, before deciding that CloudStack best suited our needs.

The installation of CloudStack can be quite challenging because you need a good understanding of the product before starting.

This solution rates a seven out of ten.

We are working as SRE, so you can consider us as an admin or consultant role because we are the ones who manage the complete cloud infrastructure. Let me introduce myself; I'm Tarun, and I'm working at Infosys, an Indian-based company. I have a total experience of 10.5 years in the IT industry, and I have worked on multiple tools such as ITSM tools, ServiceNow, BMC Remedy, and also got a chance to work on Grafana, Splunk, Prometheus, and other monitoring tools.  I work at Infosys, managing their own cloud technology, a recently developed cloud platform.

We are responsible for managing the complete infrastructure using the CloudStack platform. Our basic day-to-day activities include managing users, dealing with VM creation issues, and checking hypervisors when users encounter problems. We have various regions, including Reno, Mesa, and Maiden, where our servers run. If there is an issue with a particular hypervisor, we receive tickets from users, check the services on the hypervisor, and troubleshoot accordingly.

We also perform patching activities during weekends, including DB patching, which requires us to drain hypervisors and take downtime for management servers. Alongside these activities, we monitor everything using Grafana, which offers dashboards and alerts for issues such as service disruptions or resource utilization increases.

The features or capabilities of CloudStack that I have found most valuable include the scaling and load balancer functionalities, which are very useful as they automatically assign VMs to the appropriate host based on current CPU or memory usage. Another great feature is its ease of use, requiring no prior cloud knowledge, making it straightforward to set up the management server along with zones, pods, and clusters. Additionally, the VM migration feature is exceptional, allowing us to drain VMs to other hypervisors efficiently if a host goes down, ensuring business continuity.

CloudStack has greatly helped integrate our existing infrastructure because we manage a massive cloud platform with thousands of VMs and around 1,200 to 1,500 hypervisors. We are planning to expand to over 5,000 hosts. Given this extensive landscape, we appreciate how smoothly CloudStack handles operations without disrupting business. We maintain four management servers along with the hypervisors, distributing them across various locations such as Maiden, Reno, and Mesa, allowing us to manage our vast VM landscape effectively.

I believe enhancements could be made in CloudStack, particularly in the management servers. I had the opportunity to collaborate with one of the Apache CloudStack developers, leading to the installation of additional agents for better information retrieval from hypervisors. We implemented configurations, ensuring consistency across hosts and VMs. We also utilized Ansible or YAML files to automate password upgrades during patching activities and database management, showcasing our customization efforts for improvement.

For future releases, it would be beneficial if Apache enhanced CloudStack by integrating alerting systems directly. We've had instances where hypervisors experienced overloads without prior alerts, so proactive alerts within the tool would better prepare us for addressing issues swiftly, preventing disruption in business operations. A portal summarizing system metrics alongside alert capabilities would greatly enhance our operational efficiency.

I have been working with CloudStack for around two and a half years now.

In terms of stability and reliability, we haven't encountered stability problems with CloudStack unless there are issues with our networking team. We did face a high-priority networking issue once, but even then, CloudStack allowed us to migrate hypervisors and VMs to other zones smoothly, which demonstrates its overall reliability.

CloudStack is highly scalable. Our previous scalability efforts went well, and we are planning to implement further scalability. It's straightforward to add hypervisors to the CloudStack environment, and we manage configurations using Puppet for seamless integration, making it easy to get the services up and running promptly.

I often communicate with CloudStack's technical support and customer service. We have a Slack channel where users raise tickets or issues, and we connect with them directly to resolve problems. We log in, share screens, and troubleshoot issues from CloudStack or via CLI before closing the tickets and informing users about the resolutions.

Positive

I did not participate in the initial setup and deployment of CloudStack. By the time I arrived, the setup was already in place; however, I contributed to upgrading from 1,000 hypervisors to around 5,000, with plans to expand further.

We use CloudStack support for multiple hypervisors. Regarding network management capabilities, I find it a good feature, particularly the network offering that allows for distributing IP addresses within clusters. If a VM is created, it gets an assigned IP address, and network traffic is managed effectively. If there's excessive traffic in one region, it automatically migrates to another. We have monitoring configured, so we're informed beforehand of network issues, giving us time to act and leverage networking capabilities to manage traffic effectively. On a scale of 1-10, I rate CloudStack a 9 out of 10.