News

Good news! We’re almost there! We’ve made significant progress on almost all aspects of our OpenStack platform and we’re nearly ready for an official launch. This week, we’ve performed one final major test and maintenance cycle on the platform in order to find and fix any remaining flaws. We’ve updated configurations and software, done some serious stress testing, and conducted disaster recovery tests. Naturally, we hoped to encounter as few weaknesses as possible.

To our OpenStack testers: We trust you’re still enjoying the test period of our OpenStack platform. We’re steadily approaching the moment the service will be able to go into production. In the meantime, if you have any questions and suggestions, please don’t hesitate to contact us. This week, starting tomorrow, we’ll be performing some major maintenance and tests on the OpenStack platform. This will surely have an impact on your instances and test environment.

We’ve just released a new beta image for Cyso OpenStack for FreeBSD 10.1-RELEASE. To give you an understanding in what went into creating this image, we’ve posted an article about it on our blog. Read it here. The image can now be found in the image library.

We’ve had a really busy period with lots of OpenStack things going on. During the past month we’ve more than doubled the amount of beta testers on our platform, upgraded the entire platform to the latest OpenStack release, have worked on more features and made a great deal of progress on accounting and administrative functionality. Putting the platform through its paces Some of our new testers on the platform have been really pushing the platform and the I/O performance; we’ve had one instance of a user copying data at a speed of over 800 MB per second.

As announced in our last mailing, we will upgrade our OpenStack platform to the latest stable version: Kilo. This upgrade will be performed tomorrow: 19 May. Please expect some downtime during the day as a result. We’ll try to keep impact to a minimum, of course. Today, we also applied patches against the recently discovered Venom vulnerability that affects the platform’s virtual machines. More information can be found at Ars Technica.

On 30 April, Kilo, the eleventh stable release of OpenStack was announced. During the design summit in Vancouver later this month, the roadmap for the next release (Liberty) will be determined. We’ll be attending as well. Here at Cyso, we’ve already started testing the update to Kilo. Our first tests have been successful and we plan to roll it out to our production environment very soon. Full Kilo release notes: https://wiki.

Due to maintenance on Neutron (OpenStack Networking) yesterday, it is possible that some instances have become unreachable. To solve this issue, the instance needs to be stopped and started again.

At the beginning of this month, we sent out the first invites and officially started our beta test period. We’re already seeing a lot of activity on the OpenStack platform and we’re getting positive feedback. We’re currently expanding the number of test accounts every week. Yesterday morning, we upgraded the platform to the latest release of OpenStack Juno (the 2014.2.2 bugfix release). We’ve done this without a notification in advance to the testers in order to see if the update procedure could be performed seamlessly without any significant impact.

First of all: thank you for your patience. We first told you that we would start our beta period for our new OpenStack services in late 2014. It’s now March 2015. There have been some delays, but the wait is nearly over. Today, we’re sending out the first invitations for our OpenStack beta trial period. We’ve had far more parties interested in test driving our OpenStack platform than we had originally anticipated, so we can’t open up the platform to everyone all at once.

It’s been a while since we last gave an update about our progress regarding OpenStack. We had hoped to be able to start the public beta before the end of 2014. Unfortunately, that turned out differently. Last time, we mentioned that the platform had been upgraded to the Juno release. We had upgraded the existing platform from the previous (Icehouse) OpenStack release. In order to further test real disaster recovery, we performed a complete reinstall of the platform next.

Last month has been a busy and fruitful month. We last mentioned that our internal beta test had started and that our team visited the OpenStack Summit in Paris. For the first time, the twice-yearly OpenStack Summit took place in Europe and attracted about 6000 participants, making it the largest summit to date. It’s an indication of the momentum that OpenStack is currently experiencing. We were impressed by the amount of people that attended and the level of both professionalism and enthusiasm of the community.

Last week, a new phase in our OpenStack project started. Instead of just the members of our project team, all people at Cyso have now been invited to test our new platform and are actively doing so. It’s the final stage before we open up the platform for external beta testers. Our new group of testers has been instructed to try all sorts of different things to break the platform that our project team might not have thought of.

Today the OpenStack 2014.1.3stable Icehouse release has been announced. We’re currently testing and developing on version 2014.1.2 and will upgrade as soon as we’ve determined the changes and possible impact to our environments. Our production installation has remained stable for the last couple of weeks in which even a hardware disruption didn’t result in any impact to services. We’ve made a lot of progress with our automation tool for rolling out OTA environments and made structural changes to the use and handling of API’s.

Last week, several configurations have been created in Razor and Puppet to roll out different development environments with specific settings. These environments are using virtual Arista switches in order to resemble eventual production as closely as possible. Among other things, during the OpenStack meetup in Amsterdam on Wednesday 3 September, we’ve been given tips on how to improve performance of the erasure coded Ceph storage, which we’re intending to try out and possibly use.

Several users have continued last week’s trying out of our SSD-equipped test environments. In addition, more progress has been made towards the look and feel and future construction of the website and online community. Amongst other things, design sketches and icons have been made for users and components.

This week, focus for testing shifted mostly towards failover and recovery mechanisms and live migration of instances. Calamari, the reporting and management platform for Ceph, is producing the first graphs on usage and availability as well. Our colleagues are receiving fully virtualized test instances of the platform for them to use for providing feedback on stability, ease of use and performance. Other environments have been created for developing and testing of the APIs for the different teams within the project.

We’re continuing performance testing and benchmarking the pre-production platform, which has also been opened for individual (internal) beta test accounts. More work has been done on debugging some open issues regarding the Neutron networking component. The public API has been tested for automatic starting of images and instances. More preliminary sketches have been made of the logo.

Work is ongoing for testing performance and streamlining the installation of the platform using Puppet. We’re testing the erasure coded Ceph storage pools and different OS images (Ubuntu, CentOS, CoreOS and Windows 2012 for now). First work has been done on the final branding and a logo of our service and the user community.