Executive Summary

We’d like to thank all of you who shared your perspective on OpenNebula as part of our 2018 Architecture Survey. This is the fourth architectural survey of OpenNebula since 2012, and the results were collected during the period of December 4, 2018 through January 11, 2019. Your participation here is fundamental to our strategic focus to best provide features and support that align with the infrastructures platforms and configurations demanded by you.

We have only included in the analysis the respondents that are using OpenNebula 5.x (latest series) and who we deem reliable because they have provided identification details that allow us to verify the answers of the survey. This is important given that our main aim is to have accurate and useful information about OpenNebula deployments. This survey is not a market survey and does not express all OpenNebula deployments worldwide.

The data provided helps to shed light on how OpenNebula is being used by the community, as well as providing some indicators of where to aim for the future. In comparing to our previous survey taken in 2015, there are some other notable findings:

  • The types of workloads handled by OpenNebula is evolving, with a growing proportion (85%) of them being run in a Production environment, while also having others running in Dev/Test (73%) and in a Proof of Concept mode (29%). OpenNebula shows its increasing maturity, compared with 73% of deployments in production reported in our previous survey.
  • A few other indicators of its developing stability are both the Number of Users supported by OpenNebula clouds, as well as the Number of Nodes.  For each of these metrics, there has been steady growth across the board.
    • The growth in user count for OpenNebula clouds has progressively increased with now close to 20% servicing more than 1,000 users.  Smaller OpenNebula clouds, of 100 users or less, now constitute 45%, which is down from 70% in 2015.
    • And “node count” has seen steady growth as well, with 77% of cloud environments having more than 10 nodes, (up from 56% in 2015). Similarly, we see that 29% of cloud environments have more than 100 nodes, where in 2015 that figure was 20%.
  • Usage of OpenNebula in Commercial and Industrial enterprises continues to grow.  Its employment in organizations across various industries shows a growing distribution with increasing usage in IT, Telecommunications and Internet, and Hosting and MSP’s, and Media and Gaming sectors – mounting to 75% of total usage.  And while still an important sector of OpenNebula usage, Academia and Research constitutes a smaller percentage of the overall – now 14%, down from 25% in our last survey.
  • Hybrid cloud usage continues to grow, with an increasing percentage of OpenNebula integration with AWS (39%), as well as with Microsoft Azure (22%).
  • With the advent of Configuration Management and Remote Execution tools, we see that a hefty segment of OpenNebula users (73%) are taking advantage of these to introduce automation to their environments.
  • A growing usage of CentOS as an Operating System has reached 50% across our survey participants, while usage of both Ubuntu and Debian have remained fairly steady at 42% and 22% respectively.

And OpenNebula continues in its aim to be the simplest and most flexible open source solution for private cloud and data center virtualization management.  Our user community rates it highly because of its Simplicity (83%), its Openness (69%) and Flexibility (56%), as well as its Vendor-neutrality (52%).

Thank you again for participating in our survey! Below you can review the detailed results.

A.  About the Organization

This year we broke out the demographics across multiple categories, with “Information Technology”, “Cloud Hosting and MSP’s”, and “Telco and Internet” companies assuming large portions of the demographics – but those added together with “Media and Gaming” and “Web, SaaS, and eCommerce” total up to a growing 75% of Industrial/Commercial users. 14% of users work in Science, R&D, and Academia, while just under 11% work for Government and Non-Profits.

Type of Organization

 

Usage by larger companies has steadily grown over time, now reaching 20% of companies with 5,000 employees or more, compared to 13% from our last survey.  While smaller companies – (of 500 employees or less) – continue to be avid users of OpenNebula, that demographic constitutes a smaller base of 58%, compared to 64% from our last survey.

 

Size of Organization (# of employees)

 

48% of deployments are located in Europe or in Russia, a slight shift downward from 50% in our last survey. We also see a small tick upward in growth within North America to 33% from 30%. These two geographic regions continue to lead the usage of OpenNebula by a large margin.

Geographic Region

 

B. About Cloud Usage

The usage of OpenNebula for Production workloads has seen steady growth to 85%, from 73% in 2015. Its use for Development and Testing has remained firm at 73%, while there is significant Proof of Concept work being done, as well (19%).

Type of Workload (allows for multiple selections)

 

It is logical that the “On-premise private cloud” remains the most common type of cloud being built with OpenNebula, with 78% of respondents confirming it being part of their cloud infrastructure.  However, again the widening types of clouds being created is evident with 54% of users venturing into “Hybrid private clouds”, 55% creating “Federated private clouds”, and 17% and 26% creating “Hosted private clouds” and “Distributed private clouds”, respectively.  Even 8% are beginning to venture in creating “Edge private clouds”. With the imminent release of v.5.8 Edge, it will be interesting to see how that last figure grows.

Type of Cloud Architecture (allows for multiple selections)

 

And from a use case perspective, 71% of organizations use OpenNebula for “Data Center Virtualization management”, and 48% use it to establish “Public Clouds, VPS and MSPs”.  39% of users take advantage of OpenNebula to create cloud environments on top of their VMware infrastructure, and 37% are laying the groundwork for Enterprise clouds.

Type of Use Case (allows for multiple selections)

 

Since 2015, the number of users in most OpenNebula clouds has seen significant growth.  Clouds with more than 1,000 users have reached 19% from 8%, while the smaller ones of 100 users or less now constitute 45%, from 70% in 2015.

Number of Users

 

Another new question we introduced this year had to do with understanding with which other Cloud providers users interact.  Amazon Web Services (AWS) is, by far, the most common at 61%, with Microsoft Azure following at 35%.

Interaction with Other Cloud providers (allows for multiple selections)

 

C. About Cloud Configuration

59% of OpenNebula environments are “federated”, meaning that they have more than a single zone, and 9% are running more than 10 different zones.  This is up from 51% and 5%, respectively, from the last time we collected data.

Number of Zones (OpenNebula instances)

 

Node count – another metric in measuring cloud size – has steadily increased, as well.  77% of cloud environments have more than 10 nodes, which is a significant increase from 2015 when this measure was 56%.  And currently 29% of cloud environments have more than 100 nodes, where in 2015 that figure was 20%.

Number of Nodes

 

KVM hypervisors remain the most commonly used hypervisors in OpenNebula environments. Yet, usage with VMware hypervisor continues to be a very solid use-case. The percentage of KVM users has stayed fairly steady, at 75% compared to 73% previously.  And usage with VMware hypervisors has increased slightly, at 39% compared to 37%.

Hypervisor usage (allows for multiple selections)

 

56% of users have implemented some form of a “hybrid cloud” in their environment. And the most popular Public cloud providers used in tandem with OpenNebula are Amazon Web Services (AWS) at 39% and Microsoft Azure at 22%.  These two were most popular in our last survey, as well, but have increased slightly, where they were 30% and 16% respectively.

Public Clouds used for Hybrid (allows for multiple selections)

 

Shared datastores at 53% and Ceph at 40% remain the most widely used storage solutions in open environments – while Shared datastore usage dropping slightly from 60%, Ceph has remained the same at 40%. VMware FS at 38% is used in VMware-based deployments, mainly through vCenter, and dropped slightly from 40%.

Storage configurations (allows for multiple selections)

 

The most common network configuration is still the Standard Bridged network configuration (40%). 802.1Q VLAN (31%) and Open vSwitch (25%) remain widely-used choices as well, while there has been some movement to VXLAN and OpenSwitch VXLAN with 14% and 10% respectively.  And for VMware deployments, we see a steady usage of VMware networking (37%), primarily used through vCenter.

Networking configurations (allows for multiple selections)

 

Authentication practices, while shifting slightly to using external authentication systems, have remained fairly static.  The majority of deployments use the built-in User/Password authentication (68%), and LDAP, and Active Directory have gained slightly more traction since the last survey with 22%, and 21% usage respectively. SSH, while widely used (42%) has seen a slight drop from 50%.

Authentication configurations (allows for multiple selections)

 

CentOS and Ubuntu are still the most popular Linux distributions for creating OpenNebula clouds with usage at 48% and 41% respectively, which is a slight upward movement from 44% and 40%.  Debian has remained steady at 22% usage.

Operating Systems (allows for multiple selections)

 

A new question was introduced to understand what types of configuration management systems are employed to take advantage of automation.  The most common platform is Ansible, with 50% usage. Puppet is another popularly used tool, at 29%. There are a few other tools used with some regularity, like Chef (8%) and SaltStack (6%). And 27% of respondents state that they do not use any configuration management system at all.

Configuration Management (allows for multiple selections)

 

Similar to the types of configuration management systems, we asked which tools are used to deploy OpenNebula, and the same set of tools lined up in a similar fashion. Ansible is the most-commonly used with 40% usage, and Puppet is used by 21%.  Tools like Chef (5%) and SaltStack (4%) have a small user base, while 41% of respondents state that they do not use any deployment tools.

Deployment tools (allows for multiple selections)

 

Another question asked inquired about Container or PaaS tools being used to manage applications.  While a large part of the respondents (48%) state that they are not yet using these types of tools, for those that are using these types of tools, the most commonly used is Kubernetes with 25% usage. Docker Swarm (14%) and OpenShift (10%) follow in popularity, while 9% of respondents state they are building their own solutions.

Container / PaaS for Application Management (allows for multiple selections)

 

When asking about which Advanced Components are used or planning to be used, a large selection of components were confirmed.  The most commonly selected was “High Availability” at 65%. Other popular components are “Application Containerization” (47%), “Data Center Federation” (43%), and “OneFlow” (42%).  A few others are listed, while 14% state that they are not using or interested in using any advanced components, just yet.

Advanced Components (allows for multiple selections)

 

From a provisioning interface perspective, a large majority of OpenNebula users are taking advantage of the Sunstone interface (82% of users). The CLI and API both have a fairly extensive usage across the community with 41% and 39% respectively, and the Cloud View (Self-Service portal) has 27% of usage.

Provisioning Interfaces (allows for multiple selections)

 

We inquired about which OpenNebula API’s are being used or planned to be used in the near future, and curiously, a very large constituent (45%) has shown interest in being able to soon utilize Python bindings when v.5.8 is released. Of other API’s currently available at the time of the survey, use or intended use of Ruby bindings is common with 26%, while JAVA and Go bindings follow with 16% and 13% respectively.  29% of users state no current interest in using OpenNebula API’s.

OpenNebula API’s (allows for multiple selections)

 

And lastly, as seems to be the case for many years running, “Stability”, “Flexibility”, and “Openness” continue to be the top reasons why users choose OpenNebula.

Why OpenNebula? (allows for multiple selections)

 

Thank you again!

Stay connected!

 

Packet has announced its Edge Alliance Program, in which OpenNebula is one of the Initial Program participants.  This collaboration has been picking up steam over the past year, as focus is taking shape on Edge computing, and both platforms see a natural synergy to provide innovative solutions.  OpenNebula is just minutes away from its new version release of 5.8 “Edge”, which among other edge-focused capabilities, like providing native support for lightweight LXD containers and Automatic NIC selection, offers the ability to use bare metal providers (like Packet) to build remote clusters and to easily create Disaggregated Data Center environments along the Edge.

Packet’s Edge Alliance Program is a bold step toward encouraging innovation and providing “free access to edge computing building blocks”.  The newly announced availability of two edge sites in Chicago (IL), and a site separately deployed near Gillette Stadium in Foxborough (MA) is just the beginning of their goal to launch 15 separate site locations in 2019. With the perfect timing of OpenNebula’s 5.8 Edge release, it will become second-nature to provision Packet bare-metal resources within your cloud.

Check out more details about Packet’s Edge Alliance Program, and get engaged!

And read here for more details on OpenNebula’s partnership with Packet.

OpenNebula-based Edge Platform to be presented in 2019 Mobile World Congress

With OpenNebula as a core component, CORD (Central Office Re-architectured as a Data Center) will be featured in Telefónica’s Edge Computing demos at the Mobile World Congress in Barcelona, Spain from February 25-28. Stop by Telefonica’s booth (Hall 3, Stand 3K31) to see the new generation of Central Offices that are fully IPv6 compliant and allow for the deployment of programmable services rather than the traditional black box solutions provided by proprietary solutions.

Telefónica’s CORD prototype aims to meet low-latency demands of the emerging Internet of Things ecosystem and to virtualize the access network and give third-party IoT application developers and content providers cloud-computing capabilities at the network edge.

You can find more details surrounding the solution in this Open CORD blog.
Below are some video presentations given by Telefónica on how OpenNebula forms a key element of their innovative solution:

v.5.8 Release Candidate is available!

OpenNebula v.5.8 “Edge” is just about ready!  The Release Candidate is now available for download, which includes several bug fixes.

A huge thanks goes out to the User Community, as several of the bug fixes were found and brought to our attention by you!

For the latest details about what you can expect from v.5.8 Edge, read about it below:

Relevant Links

Beta version 2 is ready!

You will have seen our communication a few weeks ago announcing the exciting release of OpenNebula v.5.8 “Edge” – Beta version, which has been available for download.  It was a first look at the exciting new capabilities on which we have been working with a focused attention on providing edge computing capabilities in OpenNebula, as well as bringing broad enhancements recommended by our User Community.

Now, we are excited to announce the release of our Beta Version 2.  We got some great feedback from the Community over the past few weeks. While we have addressed some code fixes, we also introduced a few additional features:

  • We have added a new option for searching and filtering VM’s.
  • From a vCenter perspective, the ability to migrate datastores and clusters is now operational.
  • And you can see below the template for LXD containers!!

Once again, here’s a reminder that this is a beta release.  Its aim is for testers and developers to try the new features, and to send more feedback for the final release. Also note that being a beta, there is no migration path from the previous stable version (5.6.1) nor a migration path to the final stable version (5.8.0). A list of open issues can be found in the GitHub development portal.

Relevant Links

 

Call for Translations – extended until February 4, 2019!

We are tying up the last details for the OpenNebula 5.8 “Edge” version release.  However, before we do so, we need some help with some final translations for the Sunstone GUI portal.  Whatever help you can provide would be great!  Remember, you can translate one string at a time.

The following languages are included in current version 5.6, and the highlighted  ones require additional translations in order to be included for version 5.8:

English
* Catalan
Chinese
Czech
Dutch
* Danish
French
* German
* Greek
Italian
* Japanese
* Lithuanian
* Persian
* Polish
Portuguese (Brazil)
* Portuguese (Portugal)
Turkish
Russian
Slovak
Spanish

We’ll close the Call for Translations on Monday, Feb. 4.  Thank you for this one last push!

Closing out 2018…and welcoming in 2019!!

2018 has been another exciting year for OpenNebula. It has brought continued developments and advancements in the OpenNebula product capabilities. At the same time, we’ve seen a fervor and a steady commitment by the User Community which continues to bring unmatched value. At OpenNebula Systems, we have our sights set on continued improvement for 2019, and we are excited about several promising, emerging developments. But again, one of the key dynamics of the project is that we wouldn’t be able to grow without you.

For that reason, one of the recent developments in the community has been our request for your participation in our 2018 User Survey.  This is a simple vehicle to allow us to learn about the use cases, platforms, and overarching technical needs of the OpenNebula User Community.  We look to remain in synch with your needs, and to develop alongside with you. Fill it out, and share your thoughts!

Speaking of developments…

Among the various version releases this year, we released version 5.6 “Blue Flash” with a huge set of improvements both at the core level, as well as for vCenter integration. And from there, we have jumped right into focused development on the upcoming version 5.8.  In it, we have been working on many different features – a long-awaited support for LXD containers, being one of them.

2018 has seen a certain dedicated focus on the emerging developments surrounding Edge Computing, and while we have been working closely with customers and partners, learning the details of evolving use cases, we have also made developments around integrations with OpenNebula along “the edge”.  Earlier this year, we released an initial prototype of “oneProvision”, allowing users to provision and deploy bare-metal resources directly within an OpenNebula cluster.  Upcoming development of oneProvision will include being able to deploy not only one host, but a cluster of hosts.  At the same time, we partnered with Packet to demonstrate our continued focus in bringing capabilities to the edge.

Recent releases of new capabilities like miniONE and VirtualNetwork Scheduler, and the Image Converter to/from VMDK and QCOW2 all demonstrate our driven effort to making OpenNebula the easiest-to-use platform out there.

The “Calendar of Events”

In 2018, we held several OpenNebula TechDay events throughout Europe – in Sofia hosted by StorPool, in Barcelona hosted by CSUC, and in Frankfurt hosted by LINBIT – and in the US – in Santa Clara, CA hosted by Hitachi Vantara and in Cambridge, MA hosted by OpenNebula Systems.  We also held our OpenNebulaConf in Amsterdam.  We thank our sponsors and hosts for collaborating to put these events together.

Events schedule in 2019

The lineup for OpenNebula TechDays for this coming year will tentatively be in the following locations, with dates and details to be determined:

  • Frankfurt
  • Barcelona
  • Vienna
  • Sofia
  • Boston

And plan to attend our 2019 OpenNebulaConf in Barcelona on October 21-22, 2019.

Great support from the Community

Lastly, as we continuously try to make clear, the OpenNebula project would not have the vitality nor the reach it has if it weren’t for our dedicated User Community.  We’ve seen a continued growth of OpenNebula Champions.  Throughout the year, users have taken the time to publish tutorials like these from Pandora FMSVirtuozzo 7, and CSUC.  Our OpenNebula Blog has been used by many from the Community to publish share insight and experiences. This year, we also created our Partner Ecosystem, another instrument to show and share integrations between ONE and other great technologies.

This has been an exciting year for OpenNebula! We give you our utmost thanks, and we look forward to our collaboration going into 2019!!

Stay Connected!

The Scientific IT Services (SIS) of ETH Zurich offers  scientific computing, research data management and analysis support, as well as software engineering expertise to ETH researchers.

To support the personalized health research community, the SIS built and actively develops further “Leonhard Med”: a secure and powerful high-performance platform designed for computing, storage, management, interoperability and controlled sharing of confidential research data (e.g., biomedical patient data). Leonhard Med is operated by the Scientific IT Services (SIS) of ETH Zurich and it is part of the emerging BioMedIT national network whose role is to provide secure and interoperable data and computing infrastructures for research projects in the Swiss personalized health programs

While being in production use since beginning of 2018, Leonhard Med must be constantly developed further to keep up with new and changing requirements within a rapidly, evolving scientific environment. For example, our customers needed additional services that could not be hosted on a regular HPC infrastructure (e.g., databases, terminal servers, webapps or data management applications). This brought us to the idea of providing a cloud solution. We had some previous experience running vCloud Director (VMware) and we also had a close look to OpenStack but both came with a high price tag either in terms of license costs or manpower. Luckily one of our consultants introduced OpenNebula to us and after a few weeks of testing we fell in love with it. It met all our requirements and we found it quite intuitive and easy to maintain and support. We were actually looking for a lightweight but powerful product that is easy to maintain with few IT personnel resources and on the other hand we were aware of the challenges lying ahead of us when integrating OpenNebula into the secure environment of Leonhard Med.

We began deploying and integrating OpenNebula almost 4 months ago, using 2 physical hosts from the cluster (new hardware) and set-up the OpenNebula and 2x KVM nodes on them. We now have a fully functional and productive installation ready to serve our consumer needs and we achieved this with only a few sysadmins working on the project part time time over the four months. Our private cloud running OpenNebula sits in a restricted zone without Internet access. The access is done via proxy servers using 2 factor authentication and Sunstone is only reachable via socks proxy. For reproducibility purposes, the installation and all processes running inside the cloud has been automated with Ansible.

Challenges: We did face a couple of challenges during installation and later on during the upgrade to v5.6. For example, we had to search for a couple of ruby gems, built rpms and move them into our secured environment. These were mostly related to our network security restrictions. Nevertheless, as a “nice to have” I’d include all dependencies required during installation or upgrade within OpenNebula’s repository for RH/CentOS platforms.

Just a few minutes of your time…

As we continue to focus on improvements of OpenNebula, we need direction from you, the User Community.

Please take a few minutes to fill out this OpenNebula Survey 2018 – to help us understand how you are using OpenNebula, and what you need going forward.  All information collected is confidential, and will not be shared.

Many, many thanks!

Our newsletter contains the highlights of the OpenNebula project and its Community throughout the month.

Technology

There are a lot of OpenNebula features currently being worked on that deserve some attention:

  • We are working on an upcoming feature with the aim to simplify the management of VM templates that can be deployed on multiple clusters – by creating an automated selection process for the VM networks.  Check out the recent post.
  • Migrating workloads both to and from KVM to VMware hypervisors will soon be as simple as bread and butter for breakfast. Check out the recent post.
  • We are also working on a “self-provisioning” method for Virtual Networks.  No longer will Virtual Networks be created only by cloud administrators, but rather, end-users can be given the ability to make changes at the logic level, like changes to IP ranges, to the DNS server, etc.

Keep an eye out for the upcoming version 5.8!

Community

November has been an exciting month for the User Community.

Outreach

  • We currently have our 2019 OpenNebula TechDay Call for Hosts open.  Take a look at your calendars, and think about planning a TechDay of your own!
  • We want your feedback!!
    • In the coming weeks, we are going to be sending out an OpenNebula survey, with the intention of learning a bit about how you are using OpenNebula.  Please plan to take the time to fill it out, as its purpose is for us to be able to serve you better!
    • If you attended the OpenNebulaConf in Amsterdam, and you haven’t submitted your feedback survey, please do so, and let us know what you think!

Let’s welcome in December!

Stay Connected!