OpenStack Summit Boston Report
The path to a more agile OpenStack


Japanese version is available here

On the second week of May, just after the end of the Japanese Golden Week, I went to Boston for 2017’s first OpenStack Summit. Even though the number of attendees has slightly decreased over the years, it still managed to gather more than 5,000 engineers from all over the world, consolidating OpenStack as the second most popular open source project, only after the Linux Kernel.

With the maturity that the project has gained over the years, the focus has been reshaped: from fast growth to complexity reduction; from addition of new features to selective creation of groups dedicated to smaller but very focused projects. I find this change of focus similar in many ways to the evolution of the 802.11 standard practices.

Regarding the sessions, to cite a few, there were interesting discussions about Ceph (and how it can work with Manila at CERN [1]), neutron[2], cockroach DB [3], OVS, Barbican, Magnum, etc. The U.S. army had a session about how their soldiers are learning cyber-security [4]. Kubernetes and NFV had special sessions this summit, which will be discussed in a different article.

In the public cloud space, Deutsche Telekom presented its OpenStack based cloud [5], which has data centers all around the world and intends to become a serious alternative to the giants of the sector when it comes to hybrid clouds. Their reasoning, which sounds quite compelling, is that companies that want to use a private OpenStack cloud together with a public cloud for easy adaptability may prefer to use OpenStack for that purpose as well. That would allow for a uniform code that only needs to be written once and works everywhere in their hybrid cloud. Time will tell how their efforts pan out. Telefónica is also implementing a similar solution, but while Deutsche Telekom aims to grow in Europe and Asia, Telefónica is primarily aiming for the Americas market.

Machine learning was also touched in a few sessions. In particular, Brocade’s David Meyer gave a very inspiring presentation about the application of machine learning to network configuration [6]. This is a very promising application of the technology, but at the moment we don’t even have a network model that would work for machine learning with the currently available technologies. He invited everybody to join the effort and dig a bit deeper on what machine learning really entails.

One area that seems to be quickly leveraging the power of machine learning is the monitoring sector. Dynatrace, Datadog, Loom… they all use machine learning to detect atypical network behavior, and even identify the underlying cause. Datadog’s application is smart enought to sieve out temporary spikes (outliers), not modifying the “expected behavior” by trends that last less than a week. Loom even uses machine learning to search google and their own databases to offer the user a possible solution.

Edward Snowden gave the last keynote in an interview format with OpenStack Foundation’s COO Mark Collier [7]. He was there to remind us that our purpose is to create systems and software that work for the people, and not the other way around. It is a very simple and clear message that we would do well to keep in mind, as it may help shape some of the decisions we take in our daily jobs when we are not sure about which path to take.

Álvaro Ruiz-DelgadoITOCHU Techno-Solutions Corporation

With a background as a researcher, he worked as a 3GPP delegate for the development of the LTE standard, holding numerous patents especially in the area of MIMO communications.
Joined CTC in 2016 to leverage that knowledge helping real customers address real needs through the use of the latest techniques and technologies, going from OpenStack all the way up to the desired level of CI/CD.